Paul Deraval writes about applying AI agents to sales and business operations, with a focus on using company data to guide recommendations and deploying agents safely.
Paul Deraval is most useful for concrete examples of AI agents in sales and business operations.
Best for
A B2B SaaS media analytics campaign for explainable agent coaching and governed skills.
Not for
A consumer brand awareness campaign seeking broad reach and lifestyle content.
Key takeaways
An internal NinjaCat sales agent pairs follow-up suggestions with sales principles and account context.
An agent found a reported $750,000 retail order opportunity.
Enterprise agents need tightly scoped, composable skills, not unrestricted access.
Check first: Confirm audience composition source and recent results for comparable topics.
Main ideas
Agents should explain why they recommend something, not just issue the action · The durable asset is the data layer, not the model · Enterprise agents need tightly scoped, composable skills
Paul Deraval writes about putting AI agents to work in sales and business operations.
Their posts focus on what makes agents useful in practice: company data that supplies context, recommendations that explain their reasoning, and permissions that keep actions within safe limits.
In one example, an internal NinjaCat sales agent pairs follow-up suggestions with the sales principles and account context behind them.
Other posts describe an agent finding a reported $750,000 retail order opportunity and a live agent build for media-company operations.
Follow their work for concrete examples of how to move AI from a feature demo toward informed business decisions.
Deraval connects agent design to the conditions that make it work inside an organization.
They argue that durable value lies less in easily copied features than in account history, permissions, audit trails and other business context.
They pair that data foundation with two practical design principles: explain recommendations so people can learn from them, and compose narrowly scoped skills rather than give an agent unrestricted access.
Their posts also show the limits and tensions: SMB reporting can scale through templates but lose a tailored story, while enterprise reporting can be more personal but is harder to scale; powerful agents can act quickly, but security and trust constrain their permissions.
This combination of operational examples and design tradeoffs makes their work useful to readers assessing enterprise AI adoption, not just following model launches.
Paul Deraval argues that AI agents derive their value from the data and context beneath them and from how deliberately they are designed.
Read more
Across their posts, they argue that agents should explain their reasoning so users learn while they work, that a company's systems of record matter more than model or feature parity, and that enterprises need small, governed agent capabilities that can be chained rather than unrestricted computer access. They judge agents by the commercial action they trigger and treat personalizing reporting at scale as an unsolved problem rather than a feature win.
Agents should explain why they recommend something, not just issue the action
Deraval treats explanation as a design decision that costs almost nothing and changes the result.
Read more
A black-box agent that hands over an answer can erode users' skills, a risk they call task amnesia.
An agent that justifies each recommendation like a coach turns the interaction into a training loop: users move faster and absorb the reasoning behind the move.
In their framing, the outcome is a choice organizations make, not a fixed consequence of the technology.
NinjaCat's internal SDR Sidekick helps sales development reps decide whom to follow up with and supplies context, history, and suggested language.
It never recommends a next action without explaining the principle behind it, the sales methodology it draws on, and why it fits the situation.
Deraval says a week of daily use gives a rep more coaching surface area than a monthly call with a sales leader.
The durable asset is the data layer, not the model
Deraval argues that two years of debate about which model wins misses where value accumulates.
Read more
An agent can replicate a software feature quickly, but not years of account history, permission structures, workflow audit trails, and institutional memory.
They expect software to split into systems of record that agents query and interfaces that agents bypass entirely.
Companies that offer headless, consumption-based access—so agents read the record instead of people navigating a screen—come out strongest.
Deraval cites a CNBC report that Salesforce ingested 104 trillion records in a quarter, that Atlassian's Teamwork Graph holds 100 billion objects and connections, and that Workday says AI drove more than 25% of new contract value.
They note that these are data companies rather than AI companies.
At the Lead-to-Cash-Bash conference, they built an agent live on stage with NinjaCat customers Brian Kennett of The Minnesota Star Tribune and Shajan Thomas of Cox Media Group, attributing the speed to a solid data foundation.
Enterprise agents need tightly scoped, composable skills
Deraval separates what is technically possible from what an enterprise can responsibly run.
Read more
Unrestricted computer access produces impressive demos and unacceptable risks: leaked credentials, financial liability, damaged customer trust, and regulatory exposure.
They propose small, governed capabilities that can be chained together, so an agent that reads email cannot also send mail to anyone.
Their summary: big vision, tiny iteration.
After OpenClaw, also called Clawdbot, demonstrated what can happen when AI gets unrestricted computer access, Deraval bought a Mac Mini and called the technology remarkable. They pointed to leaked credentials from prominent people and an agent that found its owner's API keys and called them in the morning.
Deraval notes that small businesses may accept risks that enterprises cannot, making scope a deliberate tradeoff rather than a universal rule.
Deraval frames agent value in commercial outcomes rather than speed or novelty, treating trust as the bottleneck between insight and action.
Read more
Their recurring question: will someone act on what an agent finds?
A home equipment brand piloted an agent named Parker during a proof of concept.
It analyzed omnichannel sales and point-of-sale data, finding high-demand ZIP codes with zero inventory in top snow product categories during a record snow year.
The brand passed the finding to a big box retailer, which placed a $750,000 order—all before the AI agents contract was signed.
Deraval asks whether a reader would trust an agent enough to act on their next $750K opportunity.
Personalizing reporting at scale is the open problem
Deraval contrasts two reporting worlds.
Read more
Agencies serving thousands of small-business clients are pushed toward templated, cookie-cutter reports.
Teams serving big brands have room to tune the narrative, which is why their reporting often looks better.
They see scalable personalization—preserving narrative quality while serving many accounts—as the interesting question, not a solved problem.
Deraval states the tradeoff directly: an agency with thousands of clients cares about scalability, while a big-brand team can say it does not care about scalability and only wants to nail the story.
A related post argues that small-business software was already racing to zero before AI and that anyone can now build their own tool, using a metronome as a cheap example. That raises the question of what zero looks like at the enterprise level and how AI helps businesses find the next intelligent action at scale.
Shows how AI can support junior sales staff without eroding their skills, and what explainable agent coaching looks like in practice.
Read more
Paul Deraval describes an internal NinjaCat agent called SDR Sidekick, built to help sales development reps decide who to follow up with.
The agent never recommends a next action without explaining the principle behind it, including the sales methodology and specific context.
After a week of daily use, reps have absorbed more coaching than a monthly call with a sales leader would provide.
The post contrasts this with black-box agents that hand over answers and create task amnesia.
The design choice costs little: asking the agent to justify its recommendation as a coach would turns a shortcut into a training loop.
For brands planning a campaign
For a B2B SaaS media analytics platform selling contextual AI recommendations, this supports a product story built on explainability and skill development, not just speed. Every recommendation should include a reason, methodology, and contextual brief that sales teams can learn from.
Shows what makes enterprise AI defensible when model features are easy to copy, and puts the data foundation before agent design.
Read more
Paul Deraval argues the AI software debate misses that data, not model features, is the durable asset.
The post cites a recent CNBC report that Salesforce ingested 104 trillion records in a quarter, Atlassian's Teamwork Graph holds 100 billion objects and connections, and Workday says AI drove more than 25% of new contract value.
An agent can replicate a feature in an afternoon, Deraval says, but cannot replicate years of account history, permission structures, workflow audit trails, and institutional memory.
Software splits into systems of record that agents depend on and interfaces that agents bypass.
The future is a data layer plus an agentic layer, with headless, consumption-based access where agents query the record rather than humans navigate the interface.
For brands planning a campaign
For a media analytics platform assessing agent readiness, inventory the account history, permissions, and audit trails agents would need. The platform can be positioned as the system of record agents query, not just a dashboard with AI features.
Explains why enterprise agent deployments cannot copy consumer-grade automation, and outlines a governance model of scoped, composable skills.
Read more
Paul Deraval describes OpenClaw, also called Clawdbot, as a demonstration of what is possible when AI gets unfettered access to a computer—and why enterprises cannot work that way.
They bought the Mac Mini and call the technology remarkable; people are building agents to automate Shopify orders.
But tokens from prominent people leaked, security is an unmitigated nightmare, and one account describes an agent finding API keys and calling its owner in the morning.
Small businesses may accept that risk for the reward; enterprises face financial liability, customer trust, and regulatory compliance.
The answer is tightly scoped capabilities: an agent that reads email should not also send email to anyone.
Build composable skills, each governed and secure, then chain them together.
For brands planning a campaign
For an AI recommendation product aimed at sales teams, this is the counterweight to explainability: without access control, explainability is not enterprise-ready. It supports a governance layer with scoped skills, permissions, and audit trails that buyers can evaluate.
Shows what a well-scoped agent proof of concept can deliver, and how cross-channel data can produce a measurable business outcome.
Read more
Paul Deraval describes a proof of concept with a large home equipment brand selling snow blowers and lawn mowers through direct-to-consumer and big box retail.
The brand deployed an agent named Parker to analyze omnichannel sales and point-of-sale data and surface high-priority gaps.
It found top snow product categories, high-demand ZIP codes, and locations with zero inventory during a record snow year.
The brand handed the insight to a big box retailer, which placed a $750,000 order.
The order came before the AI agents contract was signed.
The post closes by asking whether a brand would trust an agent enough to act on its next $750K opportunity.
For brands planning a campaign
For a media analytics platform, this is a proof-of-concept blueprint: pick a high-value operational gap, connect cross-channel data, and let the agent surface a specific action. Measure agent value in revenue or avoided loss, not just time saved.
Shows how reporting differs by client size and why personalization at scale remains an unresolved product problem.
Read more
Paul Deraval contrasts reporting for agencies with thousands of SMB clients, which is forced into templated, cookie-cutter reports, with reporting for big brands, which gets more room to tune the narrative.
Big-brand reporting often looks better because there is room to shape the story, the post says.
The real question is how to personalize reporting at scale—and that, Deraval argues, is where things get interesting.
For brands planning a campaign
For a B2B SaaS media analytics platform, this names a buyer pain: SMB agencies need scale but lose narrative quality, while big-brand teams get narrative quality but not scalable personalization. It supports a product or campaign angle around scalable personalized reporting, with the tradeoff made explicit.
Shows how AI adoption is being demonstrated to large media companies, and the role of field events and customer proof.
Read more
Paul Deraval describes the last day of Lead-to-Cash-Bash, an annual conference for revenue and operations leaders at large media companies.
Their session with NinjaCat customers Brian Kennett from The Minnesota Star Tribune and Shajan Thomas from Cox Media Group covered how their teams use AI agents to remove friction from reporting, fulfillment, and daily operations.
The post notes how quickly an AI agent comes together when the data foundation is solid; they built one live.
Outside the session, Deraval and NinjaCat colleagues spoke with leaders ready to move past experimentation and adopt AI in meaningful ways.
For brands planning a campaign
For a media analytics platform, this suggests a field-marketing and customer-proof motion: co-present with customers, build an agent live, and use the event to move buyers from experimentation to adoption. A solid data foundation is essential for a credible live demonstration.
Explains why AI changes the economics of small-business software and shifts enterprise software's value toward data and intelligent action at scale.
Read more
Paul Deraval writes that SMB software was already in a race to zero before AI; now anyone can vibe-code their own solution, with a metronome as a cheap example. They argue all software is heading toward that horizon, but the more intriguing question is what zero looks like at the enterprise level, and how AI helps businesses identify the next intelligent action at scale.
For brands planning a campaign
For a B2B SaaS media analytics platform, this is positioning context: features alone are commoditized, so the durable offer is data, context, and intelligent next actions at enterprise scale. It supports a campaign message that ties agent value to operational decisions rather than software features.
The data foundation determines what agents can do3
Personalized reporting and insights at scale2
Software economics and the data moat2
Explainable AI agents that coach, not just act1
Enterprise-safe agents built from tightly scoped, composable skills1
Turning operational data into a specific revenue action1
Live agent builds and customer proof to move past experimentation1
Explainable AI agents that coach, not just act
Deraval argues that AI agents should explain the principle behind a recommendation, not just produce the action.
Read more
In an internal sales-development example, the agent gives reps context, history, and guidance, but never recommends a next step without explaining why. They say the choice was deliberate, cheap to add to the prompt, and qualitatively different: reps get faster while absorbing sales principles instead of depending on a black box.
For brands planning a campaign
For a B2B SaaS media analytics platform building contextual recommendation agents, this is the central design angle.
A campaign should ask Deraval to show how recommendation rationale is captured, surfaced to users, and turned into coaching or onboarding.
The proof point is not just speed: users learn the decision logic while using the product.
Deraval repeatedly ties agent performance to the underlying data layer.
Read more
In one post, they cite Salesforce, Atlassian, and Workday data volumes, then argue that AI companies are really data companies: an agent can replicate a feature quickly, but it cannot replicate account history, permission structures, workflow audit trails, or institutional memory.
In another post, they describe a pilot agent that analyzed omnichannel sales and POS data to find a specific inventory gap.
At a revenue-operations conference, they describe building an agent live and say the speed came from a solid data foundation.
Start with systems of record and operational data, then add an agentic layer.
Deraval warns that software which becomes only an interface agents bypass may not create durable value.
For brands planning a campaign
This is the strongest argument for evaluating Deraval against a media analytics platform brief.
Campaign content should not lead with model features; it should show how a unified data layer, permissions, and historical context make recommendations trustworthy.
A useful deliverable would be a data-readiness checklist or architecture narrative explaining what must be in place before an agent can act.
Enterprise-safe agents built from tightly scoped, composable skills
After discussing an openly available agent that showed what is possible when AI has unfettered computer access, Deraval says enterprises cannot work that way because of leaked credentials, financial liability, customer trust, and regulatory compliance.
Read more
They propose tightly scoped capabilities: an agent that reads email should not also be able to send email to anyone. Build composable skills, govern and secure each one, then chain them together for similar capability without the same risk.
Big vision, tiny iterations: decompose agent capability into governed, chainable skills.
Deraval notes that small businesses may accept risks that enterprises cannot.
For brands planning a campaign
For a media analytics platform that handles sensitive campaign and customer data, Deraval can speak credibly about governed agent deployment. A campaign should ask for a concrete governance model: what each agent skill can access, how permissions are enforced, how skills are composed, and what security review looks like before deployment.
Deraval contrasts agency reporting for thousands of SMB clients, where templated, cookie-cutter output is often forced, with big-brand reporting, where teams can tune the narrative.
Read more
They say the interesting question is how to personalize reporting at scale. A separate post argues that software was already in a race to zero before AI, and asks what zero looks like for enterprises and how AI helps businesses identify the next intelligent action at scale.
For brands planning a campaign
For a media analytics platform serving agencies, brands, or both, this subject maps directly to scalable personalization. A campaign should ask Deraval to demonstrate how an agent can preserve narrative quality across many accounts and distinguish template automation from genuinely contextual recommendations.
Live agent builds and customer proof to move past experimentation
Deraval describes a conference session with customers from The Minnesota Star Tribune and Cox Media Group, where they walked through using AI agents to remove friction from reporting, fulfillment, and daily operations.
Read more
They highlight building an agent live on stage and say the reaction showed that this is no longer theoretical. They also note that leaders were ready to move past experimentation toward meaningful adoption.
For brands planning a campaign
This supports event and customer-proof campaign formats.
A media analytics platform can feature Deraval in webinars, conference sessions, or customer panels where a live build or customer walkthrough demonstrates adoption.
Emphasize operational friction removed and readiness, not novelty.
Deraval argues that the AI software debate has missed the obvious point: companies with durable value may be data companies, not model companies.
Read more
They describe a split between software that becomes the system of record agents depend on and software that becomes an interface agents bypass entirely. They also argue that functionality is easy to replicate, while context is where value accumulates.
For brands planning a campaign
A media analytics platform can use Deraval to frame its market position: not another AI feature layer, but a data foundation that agents and teams can trust. Campaign material should give concrete examples of the records, history, and workflow knowledge the platform holds and explain why that context is difficult to replicate.
Paul Deraval’s posts place them at NinjaCat, a marketing and media analytics platform, working on AI agents for sales and operations teams.
Read more
The posts state no other direct brand relationships; other companies appear as examples, customers, or outside data, not as Deraval’s sponsors, clients, or partners.
NinjaCat
Employment
Deraval writes from inside NinjaCat.
Read more
Deraval says the company shipped an internal agent for its sales development reps, describes "our team" running a proof of concept with a large home-equipment brand and an agent named Parker, and refers to colleagues as "my fellow NinjaCats."
Deraval also presented at a revenue and operations conference with two NinjaCat customers, walking through how their teams use AI agents for reporting, fulfillment, and day-to-day operations.
The posts show the work Deraval does there, not a stated job title, start date, or contract terms.
Paul Deraval's posts show how AI agents move from demos to governed business tools.
Read more
Start with the SDR Sidekick post for a concrete design pattern: an agent that explains the principle behind each recommendation so sales reps learn while they work.
Then read the data-foundation argument—systems of record, permissions, and history are harder to replicate than features—and the OpenClaw post on why enterprises need small, secure, chainable agent skills.
The proof-of-concept story about a $750,000 inventory insight shows how they connect agent findings to commercial action.
For adjacent perspectives, Akhilesh Perla covers the Salesforce, MuleSoft, and Informatica stack; Ahikam Kaufman applies agent governance and token economics to finance teams; Mark Bidgood focuses on CPG data definitions and ontology.
Deraval's distinct angle is the design of explainable, coaching-oriented agents and the reporting-personalization problem for agencies and brands.
Both write about enterprise AI agents that need a governed data layer and integration across systems of record.
Read more
Perla's Salesforce and Informatica posts argue that the stack's value lies in unified, trusted data and agents that can act across business workflows.
Perla writes from a Salesforce and MuleSoft implementation practice, analyzing vendor acquisitions and platform architecture. Deraval brings a product and revenue-operations perspective, using narrative examples about agent design, coaching, and reporting personalization.
Both treat enterprise agents as governed systems that need context, controls, and a data foundation.
Read more
Kaufman applies this to finance teams, covering token economics, permissions, audit trails, and adaptation without replacing existing systems.
Kaufman focuses on finance and CFO decisions, with vendor-specific framing for Safebooks. Deraval focuses on sales, media analytics, and reporting, emphasizing explainable recommendations and composable agent skills.
Both argue that AI agents fail without agreed data definitions and a trusted operational layer.
Read more
Bidgood's CPG posts describe how inconsistent product, customer, and margin definitions cause agents to scale fragmentation.
Bidgood centers on consumer goods, retail media, supply chain, and ontology across commercial functions. Deraval centers on marketing and media analytics, sales development, and agent explainability, using first-person build stories rather than hashtag-led CPG analysis.
Placement call first: Paul Deraval's posts show a small-audience practitioner profile, not a reach play.
Read more
The one measured audience figure is a LinkedIn profile showing about 5,052 followers as of 8 October 2026.
That size fits a credibility-led placement aimed at people already working on AI agents for sales and operations—a sponsored post, co-authored explainer or webinar guest slot—rather than a campaign that needs broad distribution or cheap impressions.
For a B2B SaaS media analytics platform selling explainable agent coaching into sales teams, expect influence with a narrow, on-topic group and judge the placement on engagement and lead quality, not impressions.
The topics point to sales development and revenue operations staff, leaders at large media and advertising companies, and retail, DTC and consumer equipment brand teams working with omnichannel sales data.
Two posts make that audience visible: a session delivered to revenue and ops leaders at a media-industry conference, and a proof-of-concept story about a home equipment brand using an agent to spot retail stock gaps.
These are the readers the posts address, not a verified breakdown of followers.
The posts do not measure followers' geography, age, gender, employer, job function or seniority.
If you need to know whether the audience is director-level sales operations rather than individual contributors or analysts, ask whether Deraval has audience survey results or another documented source for that attribute, and request it alongside comparable-topic performance results.
Selected examples include a detailed explanation of how an internal sales agent at NinjaCat justifies each recommended next action, and a write-up of an agent that surfaced a $750,000 retail order opportunity during a proof of concept.
Both show the level of specificity in the content; neither is audience data.
Before committing budget, ask for recent results for comparable topics, depth and formats—for example, reach and engagement on posts about agent design, data foundations and reporting at scale, plus any sponsored placement history.
Ask for the audience composition source separately.
Different accumulated views across posts do not show that a topic or format caused the difference; only delivery results for comparable work under comparable conditions would inform the placement decision.
Paul Deraval's posts document their work at NinjaCat building and deploying AI agents: an internal sales-development agent that explains the reasoning behind each recommendation, a customer pilot in home equipment retail that surfaced $750,000 in new orders before a contract was signed, and a conference session with media-company operations leaders on removing friction from reporting and fulfillment.
Read more
The posts offer practitioner commentary and company case examples, but no sponsorship terms, rate card, media kit, or partnership inquiry route. Brands should ask directly about available formats, deliverables, pricing, approval process, and how to get in touch.
This is a strong content fit for a campaign for a B2B SaaS media analytics platform seeking AI agents that deliver contextual recommendations, explainable coaching, data-foundation readiness, governed skill composition, and scaled personalization for sales teams.
Read more
Deraval's posts address each part directly.
The clearest match is the SDR Sidekick example: an internal sales agent that never recommends a next action without explaining the principle behind it, the sales methodology it draws on, and why the move fits the context.
That is almost the exact product problem for the campaign, giving it a concrete design requirement rather than a vague claim about AI coaching.
The second match is the argument that context, account history, permissions, and workflow audit trails matter more than model features.
That supports positioning the platform's data layer as the foundation agents query.
The third is the governance model: tightly scoped, composable skills that are governed and chained together, rather than unfettered agents that create security and compliance risk.
The fourth is the reporting-at-scale problem: agencies with thousands of SMB clients are stuck with cookie-cutter reports, while big-brand teams can tune the narrative.
That maps to a platform selling contextual recommendations across many accounts.
The fifth is the live-build and customer-proof motion at a conference for revenue and operations leaders at large media companies.
The audience is small, at about 5,052 LinkedIn followers, so this is a narrow credibility placement rather than a reach play.
The posts do not mention sponsorship offerings, a rate card, a media kit, or a partnership inquiry route; ask about these before proceeding.
The best fit is a co-authored explainer, design brief, webinar, or customer panel where Deraval can discuss agent design, data readiness, governance, and personalized reporting with specificity.
Judge the placement on lead quality and engagement with sales and operations readers, not impressions.
When to choose a specialist
A materially different campaign would be a consumer brand-awareness push seeking broad reach and lifestyle content, such as a new consumer packaged goods or fashion launch targeting US adults 18 to 44 with short-form video, influencer product reviews, and a large-volume impression target.
Read more
That is a poor fit based on what Paul Deraval's posts say.
The only measured audience figure is a LinkedIn profile showing about 5,052 followers, which cannot support a mass-reach goal.
The content focuses on enterprise AI agents, sales development, data foundations, media revenue operations, reporting personalization, and software economics.
The posts do not mention consumer lifestyle content, product reviews, short-form video, beauty, fashion, food, or broad consumer audience composition.
They also do not mention sponsorship formats, deliverables, pricing, or an inquiry route; ask about these before considering a consumer campaign.
A campaign that ranks creators by follower count or requires guaranteed impressions is not supported by the posts.
A campaign seeking rapid consumer product endorsement is also a poor fit: the posts are first-person practitioner accounts and company case examples, not independent consumer recommendations.
The campaign should shift to a B2B audience or choose a different creator.
A worked campaign example
Worked campaign angle for a campaign seeking contextual recommendations, explainable coaching, data-foundation readiness, governed skill composition, and scaled personalization: explain the recommendation, then prove the data foundation.
The angle is a co-authored design brief or webinar called something like “The agent that shows its work,” built around the SDR Sidekick principle that an agent should never recommend a next action without explaining the reasoning and the sales methodology behind it.
The campaign starts with a concrete problem: sales teams get faster with AI, but junior reps can lose the ability to make decisions if the agent is a black box.
The platform’s answer is contextual recommendation with visible rationale.
The narrative then moves to the data foundation.
Before an agent can coach a sales team, the media analytics platform needs account history, permissions, workflow audit trails, and institutional memory in one queryable layer.
Deraval’s data-layer argument supplies the frame: features are easy to copy, but context is durable.
The next section covers governance.
Show which skill each agent can use, what it can read or write, how permissions are enforced, and how scoped skills are chained together so an agent that reads email cannot also send to anyone.
Then scale the example to reporting.
Use the agency-versus-big-brand contrast to show why personalization is hard at scale, and demonstrate how an agent can preserve narrative quality across thousands of SMB accounts without reverting to cookie-cutter templates.
For proof, use the Parker pilot story as an outcome-led company case example: an agent analyzed omnichannel sales and point-of-sale data, found high-demand ZIP codes with zero inventory, and a big-box retailer placed a $750,000 order before the contract was signed.
Ask for supporting evidence before presenting the outcome as independently verified.
The call to action is a data-readiness checklist and a working session with the platform’s product and sales operations leads.
A practical test for the campaign: after seeing the explainer, can a sales leader describe why an agent recommended a next action, what data it used, and what permission boundary applied?
If not, the creative has not delivered the design approach.
The posts do not mention partnership formats; negotiate directly rather than presenting an off-the-shelf sponsorship package as available.
Choose Perla for a campaign aimed at enterprise architects evaluating Salesforce and MuleSoft.
Read more
In two posts, Perla presents Salesforce as an expanding enterprise platform, arguing that data governance, integration, and agent capabilities must work together for AI to operate across business systems.
Perla is a stronger fit than Deraval for a campaign about platform architecture and integration, rather than how an agent can coach a seller through its recommendations.
Perla’s examples focus on Salesforce architecture and integration, so a sales-enablement or media-operations campaign would need to translate the argument for its audience. Deraval offers a more direct example: an SDR agent that explains the reasoning behind its suggested next step, turning the interaction into a coaching opportunity as well as a productivity aid.
Choose Kaufman when the audience is finance leaders weighing the costs and risks of putting AI into financial workflows.
Read more
Kaufman connects agent usage—model choice, token consumption, and controls—to the broader question of whether a finance team can adapt if its systems or needs change.
Kaufman is a better fit than Deraval when the campaign’s central question is financial governance or flexibility, not the quality of an agent’s recommendations.
Kaufman’s examples center on finance processes such as reconciliation, investigation, and close work.
They offer less direct material for a campaign about sales coaching or media reporting than Deraval’s SDR Sidekick and reporting examples.
A sales or media-operations brief would need to recast these finance-specific use cases.
Choose Bidgood for a consumer-goods campaign about data definitions and alignment across functions.
Read more
In the post on product definitions in CPG, Bidgood shows how different definitions across marketing, supply chain, finance, and retail can lead AI to optimize against inconsistent signals.
A second post on trusted data makes the related point that connected systems are not enough if the business does not agree on what its data means.
Bidgood is a better fit than Deraval when the buyer’s first challenge is aligning business definitions before deploying AI, especially in CPG or retail media.
Bidgood’s examples are specific to consumer goods and focus on data meaning across functions, not on how an agent explains its recommendations to a seller. For a sales-enablement brief, Deraval’s SDR Sidekick is a more direct example; adapting Bidgood’s argument would require new audience-specific proof points.
Paul Deraval's seven LinkedIn posts cover an internal sales agent that explains the reasoning behind each recommendation, the argument that durable software value sits in data and context rather than model features, enterprise agent security and scoped permissions, a proof-of-concept story about a reported $750,000 retail order, reporting personalization for agencies and brands, a media-industry conference session, and software economics.
A LinkedIn profile snapshot dated 8 October 2026 shows about 5,052 followers.
Deraval's angle sits alongside those of adjacent enterprise AI writers, and the overview, ideas, relationships, discovery, and campaign-fit sections are presented here.
Gaps that remain open: Paul Deraval's posts carry no publication dates, so changes over time cannot be placed on a verified timeline.
The posts do not mention signature books, shows, or earlier career work that might also explain searches for this name.
No sponsorship formats, media kit, pricing, or partnership inquiry route appear here, and no audience composition or performance data are available.
Only NinjaCat appears as an ongoing brand relationship; other companies named in the posts are examples, customers, or third-party data rather than Deraval's sponsors or clients.
