Juan Jaysingh writes about AI agents in customer service, focusing on how businesses can keep decisions consistent across channels by combining AI reasoning with rules and human judgment.
Juan Jaysingh is most useful for decision orchestration in customer service, keeping AI reasoning, rules, and human judgment consistent across channels.
Best for
B2B SaaS AI customer service tools; thought leadership on agentic CX decision orchestration.
Not for
B2B software vendors outside customer service seeking sponsored short-form video demos or audience composition and engagement figures.
Key takeaways
Deciding what to do is the hard part, not understanding the request.
Permissions are ahead of procedures, and a procedure without enforcement is only a request.
Complex customer service is several different problems, each needing its own control.
Check first: Confirm format, scope, and requirements, and request recent LinkedIn audience analytics.
Main ideas
Deciding what to do is the hard part, not understanding the request · Permissions are ahead of procedures, and a procedure without enforcement is only a request · Complex customer service is several different problems, each needing its own control
Juan Jaysingh writes about how businesses can use AI agents in customer service without letting decisions drift across channels.
Their work focuses on the line between what AI can interpret and what should be governed by defined rules or human judgment.
In a damaged-order example, the right response may depend on available information, item value and company policy; separate agents could otherwise approve a refund, deny it or promise a replacement.
They also examine AI assistants connected to CRM systems, arguing that permission to take an action is not the same as guidance—and enforcement—for how to take it.
Follow their work for clear, practical thinking about decision orchestration and consistent customer-service workflows.
Jaysingh makes AI governance tangible by tracing a decision through the customer experience.
Their damaged-delivery example shows why a model that understands a request still needs shared business rules: without them, different channels can give the same customer conflicting answers.
They apply the same distinction to CRM tools connected to AI assistants: permissions define what a user can access, while procedures and enforcement shape how actions are taken.
Their proposed direction is a common decision layer for many interfaces, with AI reasoning where appropriate, rules for consequential decisions and people involved when judgment is needed.
A post about complex customer service adds an important nuance: different risks call for different controls, from compliance safeguards to human intervention.
Juan Jaysingh argues that understanding what a customer wants is the easy part of AI customer service; deciding what to do consistently is harder.
Read more
Across three posts, they distinguish what a model should reason about from what must be enforced deterministically, and argue that business logic belongs outside the model, in a shared decision layer used by every channel and agent.
The priority is control: where reasoning is allowed, which decisions must follow fixed rules, and when a person steps in.
Jaysingh also makes uncertainty part of the case, listing what is known about a vendor release and stating what remains unknown.
They quote Brent Thomas of Zingtree for a specific point about models producing answers that cannot be inspected.
Deciding what to do is the hard part, not understanding the request
Jaysingh argues that AI can already read intent and produce straightforward replies.
Read more
The harder task is deciding whether a customer qualifies for a particular outcome when policy, value thresholds and missing evidence all apply.
They propose moving that business logic out of the model and into a decision orchestration layer reachable through MCP, so chat, voice, text and human agents reach the same answer.
Their point: the right answer should not depend on the channel; everyone should sing from the same sheet music.
A customer says an order arrived damaged and asks for a refund.
They have no photo evidence, the item is above a certain value, and policy favors a replacement.
If each AI agent decides alone, chat approves the refund, voice declines it, and text confirms a replacement is on its way.
Permissions are ahead of procedures, and a procedure without enforcement is only a request
Writing about Salesforce and Anthropic putting the CRM inside Claude, Jaysingh separates what is known from what remains unknown.
Read more
An admin connects Salesforce once, permissions are managed centrally, and actions route back through Salesforce so business rules are enforced.
But which rules, and how they are enforced, remain unclear.
That gap is the argument.
Permissions govern what a person may access; procedures provide rules-based guidance on what to do.
Without enforcement, a procedure is a polite request, not a control.
Jaysingh argues that the model cannot serve as the procedure layer: procedures cannot be hard-coded deterministically inside a model, and a procedure in a prompt can be interpreted differently depending on context.
Their closing instruction: verify rather than trust.
A permission is not a procedure: knowing someone may issue a credit does not tell you when to issue one instead of a replacement. Central permissions cannot enforce a rule the system was never given.
Complex customer service is several different problems, each needing its own control
In a post about an article for CRM Magazine, Jaysingh pushes back on treating complex customer service as a single problem.
Read more
They distinguish at least three forms of complexity, argue each calls for a different kind of control, and frame the real questions: where should AI reason, which decisions must follow defined rules, and when should a person step in? Their recommendation is a hybrid: give AI room to reason, enforce the rules that matter, and involve people where judgment is needed.
Jaysingh identifies three forms of complexity: a wrong answer that creates compliance risk, an incorrect approval that costs money, and a request that depends on live data from several systems, with many possible paths and little time to act.
Headless optionality is the shift worth watching, not integrations or APIs
Jaysingh argues that SaaS spent twenty years competing over who owns the screen, and that agentic headless optionality changes the equation.
Read more
They rate headless above integrations or APIs because one decision layer can serve many interfaces, and read the Salesforce–Anthropic release as a signal of where headless is heading.
One decision layer serving chat, voice, text and human agents, so the answer does not depend on the interface the customer uses.
The two posts share a tension, not a reversal.
Jaysingh is enthusiastic about headless optionality while acknowledging that the enforcement layer for procedures and rules is not yet in place.
The direction they find most interesting therefore carries the most unresolved risk.
Their market read draws on recent vendor developments, not a survey of the category.
Naming what is still unknown is part of the argument
Jaysingh lists what is known about a vendor connection, states plainly what remains unknown, and closes with an instruction to verify rather than trust.
Read more
Brent Thomas of Zingtree offers a specific point about models producing plausible answers without an inspectable reason. Jaysingh uses the quote to support moving logic out of the model, not to pad the post.
Brent Thomas, quoted by Jaysingh: AI is really good at making stuff up, or at least at taking its best guess, and it is not good at always saying that one plus one is two; sometimes one plus one is two and a half, for reasons you cannot inspect.
This is the clearest single explanation of the problem Jaysingh keeps returning to: understanding what a customer wants is not the same as deciding what to do about it.
Read more
Start here to see the distinction in action before reading anything more abstract.
Jaysingh responds to a CX Today question about what the CRM becomes when the interface is no longer the centre of the customer experience.
They answer that the CRM becomes part of the journey rather than a destination.
The piece works because it grounds the argument in one small case: a customer says an order arrived damaged and asks for a refund.
AI handles the intent and the happy-path reply easily, they write, but the harder problem is deciding whether that customer qualifies when there is no photo evidence, when the item is above a certain value, and when policy prefers a replacement over a refund.
Left to individual agents, the outcome fragments: chat approves the refund, voice declines it, and text confirms a replacement.
They quote Brent Thomas of Zingtree on models producing plausible answers that cannot be inspected, then propose moving business logic out of the model into a decision orchestration layer reachable through MCP, so every channel reads from the same sheet music.
It is a concrete argument rather than a prediction, and the damaged-order example is what makes it stick.
For brands planning a campaign
A thought leadership piece can borrow the structure directly: open with one messy customer case, show the channels disagreeing, then make the case for a shared decision layer. The refund scenario is specific enough to carry an article without further invention.
Answers the question a sceptical buyer is already asking: if the CRM is now reachable from inside a general assistant, what actually enforces the business rules?
Read more
It shows how Jaysingh separates what is known from what is still unproven.
Jaysingh examines Salesforce and Anthropic putting the CRM inside Claude, and notes the obvious objections: was Salesforce not already headless, and is Agentforce still needed?
The post is organised around the split between what is known and what nobody knows yet.
Known: an admin connects Salesforce once, permissions are managed centrally, and actions route back through Salesforce so business rules are enforced.
Unknown: which rules, and how they are enforced.
That gap is the argument.
Permissions govern what a person may touch; procedures are the rules-based guidance for how something should be done, and a procedure without enforcement is a polite request rather than a guarantee.
They point out there is nowhere inside Claude to deterministically hard-code a procedure — a prompt can be written but will be interpreted differently depending on context — and close by observing that most vendors in the space are further along on permissions than on procedures, with the warning "Don't trust, verify!"
The post rewards a reader who wants a vendor announcement taken apart rather than repeated.
For brands planning a campaign
Useful for a piece that positions a product as the enforcement layer rather than another assistant. The permissions-versus-procedures distinction gives a vendor-neutral frame that avoids criticising any single platform.
It defines the categories of difficulty and states the hybrid recommendation in a few lines, which makes it the right first stop for a reader who wants the thesis before the detail.
Jaysingh notes writing an article for CRM Magazine on where AI agents still struggle in complex customer service, and uses the post to break "complex CX" into distinct kinds of problem: a wrong answer that creates a compliance risk, an incorrect approval that costs money, and a request that depends on live data from several systems with many possible paths and little time to act.
Each, they argue, calls for a different kind of control, and the real question is where AI should reason, which decisions must follow defined rules, and when a person needs to step in.
The recommendation is a hybrid: give AI room to reason, enforce the rules that matter, and bring in people where judgment is needed.
For brands planning a campaign
The three-way breakdown of complexity gives a clean outline for a co-authored article or webinar, with each category mapped to a different control mechanism. Worth pairing with the damaged-order example from the decision orchestration post.
Decision orchestration: one answer across every channel
Jaysingh argues that CRM stops being a destination and becomes part of an agentic journey when the interface is no longer the centre of the customer experience.
Read more
Their example is a support case in which a customer reports a damaged order and asks for a refund.
AI handles intent and the happy path, but eligibility questions — no photo evidence, an item above a value threshold, or a policy that prefers replacement over refund — pose a different problem.
If each agent decides alone, chat approves the refund, voice declines it, and a text confirms a replacement is on its way.
Moving business logic out of the model and into a decision orchestration layer reachable through MCP gives the same answer across channels, whether a human or an AI agent handles the conversation.
Starts with a concrete customer-service case and follows it across three channels to show where consistency breaks.
For brands planning a campaign
A thought leadership article on decision orchestration could open with the damaged-order case, positioning the brand's product as the layer that carries one decision across chat, voice, text, and human agents. Pair Jaysingh with an internal architect if the article needs implementation detail beyond the argument.
Permissions are further along than procedures and enforcement
Writing about Salesforce and Anthropic connecting the CRM inside Claude, Jaysingh separates what's known from what's not.
Read more
An admin connects Salesforce once, permissions are managed centrally, and actions route through Salesforce to enforce business rules.
What's unknown is which rules apply and how they are enforced.
They distinguish permissions — what a person may touch — from procedures, the rules for how to act.
A procedure without enforcement is a request, not a control.
They argue that the model cannot serve as the procedure layer, not as a criticism of the model: there is no place inside a model to hard-code a procedure deterministically, and a prompt can be interpreted differently depending on context.
Their conclusion: almost every vendor in this space is further along on permissions than on procedures and enforcement.
Separates confirmed capabilities from open questions before drawing a conclusion, then generalises from one vendor release to the wider market.
For brands planning a campaign
A strong angle for the brand's article is the enforcement gap between permissions and procedures.
The post also offers a template for a credible, vendor-neutral section that names what's not yet known.
That approach suits a SaaS company selling the missing layer, provided the piece stays clear about what the product can and cannot enforce.
Complex CX means several different problems, each needing a different control
In an article for CRM Magazine, Jaysingh pushes back on treating complex customer service as one kind of problem.
Read more
They identify at least three forms of complexity: a wrong answer that creates a compliance risk, an incorrect approval that costs money, and a request that depends on live data from several systems, with many possible paths and little time to act.
Each calls for a different kind of control.
The result is a hybrid position: give AI room to reason, enforce the rules that matter, and bring in people when judgment is needed.
For brands planning a campaign
A useful structure for a co-authored piece: three named types of complexity mapped to three control choices. It lets a customer-service AI vendor show where reasoning, deterministic rules, and human escalation belong, rather than claiming AI handles everything.
Headless optionality is the interesting shift, not integrations or APIs
Across two posts, Jaysingh returns to a claim about the market: SaaS spent twenty years competing over who owns the screen, and agentic headless optionality changes that.
Read more
They rank headless above integrations and APIs because one decision layer can serve many interfaces. They read the Salesforce–Anthropic release as a signal about where headless is heading: more opportunity, more complexity, and more risk in business workflows.
Uses a vendor announcement to interpret market direction rather than review a product.
The market argument draws on two recent vendor developments, not a survey of the category. Treat it as a position to test in the article, not an established finding.
For brands planning a campaign
Frames the article around a market shift the brand can own: as interfaces multiply, consistency depends on the decision layer. A useful hook for an opening section aimed at buyers already running several channels.
Publishing the open questions instead of filling them in
Jaysingh makes the open questions visible.
Read more
They list what's known about the Salesforce–Anthropic connection, state plainly what's not known, and close with a verify-don't-trust instruction. They also quote Brent Thomas of Zingtree on models taking a best guess without an inspectable reason, using the quote to support the case for moving logic out of the model rather than to pad the post.
Quotes a named practitioner on one specific point, then returns to Jaysingh's argument.
For brands planning a campaign
A brand article could use the same structure: state what's confirmed, mark what's not, and include one outside voice for a single sharp claim. It reads as confidence rather than hedging and gives the article a natural place to invite reader responses.
The posts document one editorial placement and one quoted commentary appearance, plus analytical discussion of two vendor platforms and a source quoted from Zingtree.
Read more
Juan Jaysingh wrote an article for CRM Magazine on where AI agents struggle in complex customer service, and CX Today published their answer on what the CRM becomes when the interface is no longer the center of the customer experience.
The remaining brand references are subjects of analysis: Salesforce and its Agentforce product, Anthropic and its Claude product, and a Zingtree article by Brent Thomas that Jaysingh quotes on the limits of AI decision-making.
None of the posts states an employer, ownership stake, sponsorship, affiliate arrangement, or paid client engagement.
Ask whether any such commercial relationships exist before booking.
CRM Magazine
Disclosed partnership
Jaysingh argued in an article for CRM Magazine for a hybrid approach to complex customer service: let AI reason, enforce the rules that must hold, and bring people in when judgment is needed.
Read more
The post announces the byline but does not state whether the placement was paid, commissioned, or a standard editorial contribution. Ask about the commercial terms.
CX Today asked Jaysingh what the CRM becomes when the interface is no longer the center of the customer experience.
Read more
They answered that the CRM becomes part of the agentic journey rather than a destination. The post features a quoted commentary appearance; ask whether it was paid or sponsored.
Salesforce is the subject of Jaysingh’s analysis of the Claudeforce release, which put the CRM inside Claude.
Read more
They raise questions about which business rules are enforced and how, noting that vendors are generally further along on permissions than on procedure enforcement.
Anthropic’s Claude product now hosts Salesforce CRM workflows after the Claudeforce integration.
Read more
Jaysingh argues that a model cannot serve as the deterministic rules layer because a procedure written into a prompt can be interpreted differently depending on context.
Jaysingh quotes a Zingtree article by Brent Thomas on why AI struggles to say that one plus one is reliably two, using it to support their argument for moving business logic into a decision orchestration layer.
Juan Jaysingh writes on LinkedIn about where AI agents fit in customer service and what has to sit outside them.
Read more
Read three posts as one argument: understanding what a customer wants is easy; deciding what to do about it consistently is hard.
Start with the post on decision orchestration, which uses a damaged-order refund that chat, voice and text each resolve differently.
Then read the post on Salesforce and Anthropic putting the CRM inside Claude, which draws the line between permissions and procedures, followed by the shorter post pointing to a CRM Magazine article that separates three kinds of service complexity.
The posts link to longer bylined articles in trade publications.
Compare Jaysingh with other writers on enterprise AI adoption, agent readiness and business-process automation; three are listed below.
Perla writes about the same vendor stack and the question of what has to be true before agents can act: Salesforce, Informatica, MuleSoft and Agentforce, and the argument that governed, clean data must come before autonomous execution.
Read more
Both writers focus on the layer around the model as the place where enterprise AI succeeds or fails, and both work close to Salesforce implementations.
Perla argues from the platform’s own packaging, breaking the acquisition of Informatica into a numbered list of what the combined stack unlocks and closing with a direct invitation to get in touch.
Jaysingh argues from a single customer case and the distinction between permissions and procedures.
Their posts name what is still unknown about a release rather than what it delivers.
Readers wanting a vendor-stack view should read Perla; those wanting the enforcement gap examined should read Jaysingh.
Votek examines whether enterprise software is ready for agents to be the users, scoring twenty-two vendors on API openness, agent-native features, data portability, real-time access and ecosystem flexibility.
Read more
They also comment on consulting firms rebuilding themselves around agents. This overlaps with Jaysingh’s headless-optionality argument: which platforms let an agent act, and what changes when the interface stops being the point.
Votek uses scored comparisons across companies, with named scores and a ranked carousel, plus reported commentary from a consulting leader.
Jaysingh works through a service scenario and distinguishes between two kinds of control.
Votek answers which vendors are furthest along; Jaysingh asks where the decision itself should live.
Barwal covers the operational side of the same shift: which internal processes a mid-market business should automate first, including lead qualification and routing, tier-one support and order inquiries, and syncing data across a website, CRM, ERP, accounting and mobile app.
Read more
Jaysingh’s damaged-order refund is a support case of the kind Barwal identifies as a candidate for an agent. Both writers focus on what happens when several systems hold part of the answer.
Barwal writes prescriptive checklists for operations teams, with numbered fixes and a closing question about how many processes remain manual.
They also cover eCommerce page speed and conversion.
Jaysingh focuses on where reasoning is allowed and what must be enforced, without offering implementation steps.
Barwal is the practical starting point; Jaysingh makes the argument about control.
Juan Jaysingh’s posts are relevant to a campaign on decision orchestration in agentic customer experience, discussing consistent business rules across customer-service channels and the combination of AI reasoning, rule enforcement and human judgment.
Read more
Before selecting them, request recent LinkedIn audience analytics and, if professional roles or seniority matter to the campaign, survey results or other data covering those attributes.
Juan Jaysingh's three posts analyze AI agents in customer service, CRM decision workflows, and decision orchestration.
Read more
They don't describe partnership formats, announce a sponsorship, publish a media kit or rate card, or set out terms for working with a brand. Buyers should confirm format, scope, and requirements directly before planning around them.
Brief: a B2B SaaS company that sells AI-powered customer service tools wants a thought leadership article on decision orchestration in agentic CX, aimed at contact-centre and CX platform buyers.
Read more
This is a strong fit on subject matter.
All three posts sit directly on the argument the brief asks for.
The most useful is the decision orchestration post: it opens with a customer saying an order arrived damaged and asking for a refund, notes that AI handles the intent and the happy-path reply, then shows where the real difficulty lies — no photo evidence, an item above a value threshold, a policy that prefers replacement over refund.
Left to individual agents, chat approves the refund, voice declines it, and text confirms a replacement.
The proposed fix is to move business logic out of the model into a decision orchestration layer reachable through MCP so every channel reads from the same sheet music.
That is the brief’s thesis, already worked through with a concrete case the article can reuse.
Two other posts strengthen the fit.
The Salesforce-Anthropic piece separates what is known from what nobody knows yet, and makes the distinction a buyer can act on: permissions say what a person may touch, procedures are the rules for how, and a procedure without enforcement is a request rather than a control.
It also argues there is no place inside a model to hard-code a procedure deterministically, and closes by noting most vendors are further along on permissions than on procedures.
The CRM Magazine post breaks “complex CX” into three different problems — a wrong answer that creates a compliance risk, an incorrect approval that costs money, and a request depending on live data across several systems — and maps each to a different control, landing on a hybrid position: let AI reason, enforce the rules that matter, bring people in where judgment is needed.
The reasons to choose Jaysingh are narrow but real.
They reason from named cases rather than market mood, separate confirmed capability from open questions, and are willing to say a vendor release raises questions it does not answer.
For a brand selling the missing enforcement layer, that posture is an asset: the piece can argue for a decision layer without pretending every product claim is already proven.
Before commissioning, confirm format, scope, rights, and disclosure directly; the posts provide no partnership terms.
The market-direction claim — that headless optionality matters more than integrations or APIs — is based on two recent vendor developments, not a category survey, so the article should frame it as a position under test rather than an established finding.
When to choose a specialist
Brief: a B2B software vendor outside customer service — for example sales enablement, marketing analytics, or revenue operations — wants a sponsored short-form video series in which the creator walks through the sponsor’s product and posts demo clips, with selection judged on audience composition and engagement performance.
Read more
This is a poor fit against what the posts show, for four reasons.
First, format: all three items are written posts, two summarising bylined trade articles.
No video, audio, podcast, or live format appears, so there is nothing to judge video capability on.
Second, audience proof: the brief’s selection criterion is who watches and how they engage.
The posts contain no audience composition data or engagement figures, and the follower count attached to the account is not a verified metric.
Third, commercial terms: no sponsorship format, rate card, media kit, or disclosed sponsored content appears, so the brief cannot be scoped or priced from what is available.
Fourth, subject distance: the posts cover customer-service decisioning, CRM architecture, agentic customer experience, and enforcement of business rules.
Sales enablement or marketing analytics sits far enough away that the article-based credibility demonstrated here does not transfer without new evidence.
To assess the brief, collect a sample of any video or live appearances, current audience and engagement data for the channel, and confirmation of the formats and terms Jaysingh will consider.
If the sponsor is willing to shift to a written format and a customer-service angle, the fit picture changes and the good-fit brief above becomes the closer comparable.
A worked campaign example
Worked campaign angle for the good-fit brief.
Working title: “One Decision, Many Interfaces — Why Intent Is Solved and Decisioning Is Not.”
Where it runs: the sponsor’s blog and LinkedIn, with a short excerpt post.
Roughly 1,200 to 1,500 words.
Single byline for Juan Jaysingh, or co-bylined with a named architect from the sponsor if the piece needs implementation detail beyond the argument.
Opening: the damaged order.
A customer reports damage and asks for a refund.
State plainly what AI already does well — reads the intent, drafts the happy-path reply — then what it cannot settle alone: no photo evidence, an item above a value threshold, a policy that prefers replacement over refund.
Show the failure as three channels giving three answers: chat approves, voice declines, text promises a replacement.
Close the opening by naming the actual problem: consistency across interfaces, not comprehension.
Section one — permissions are not procedures.
Use the sponsor’s own frame here: permissions say what a person may touch; procedures are the rules for how.
A procedure without enforcement is a request, not a control.
Keep the point vendor-neutral so the piece does not read as an attack on any named platform.
Section two — three kinds of complexity, three kinds of control.
Take the three categories from the CRM Magazine post: a wrong answer that creates a compliance risk, an incorrect approval that costs money, and a request depending on live data from several systems with many possible paths and little time.
Map each to a control — reasoning, deterministic rule enforcement, human escalation.
This is where the sponsor’s product earns its place, as the layer holding one decision across chat, voice, text, and human agents.
Section three — what to ask before buying.
A short buyer’s checklist: which rules are enforced, where they live, who can change them, what happens when a rule and a model disagree, and what the escalation path is when judgment is required.
Keep the deliberate gaps visible — confirm what is known, mark what is not — rather than filling every uncertainty with a product claim.
Close: a hybrid recommendation — let AI reason, enforce the rules that matter, bring in people where judgment is needed — and a concrete next step inviting the reader to bring one decision that keeps failing so it can be traced across their own channels.
What the sponsor provides: the enforcement and orchestration detail the posts deliberately leave open, plus a reviewer for technical accuracy.
What Jaysingh provides: the argument, the case, and the buyer-facing framing.
Their published commentary is the source; the article should be original prose, not a repost.
Before writing, confirm scope, deadline, rights, review process, and disclosure directly with Jaysingh. Treat the market-direction claim about headless optionality as a position to argue in the piece, not as a surveyed finding.
Choose Perla when the article needs an enterprise-platform argument: which layers must exist before an agent can safely act on a customer-service decision, and why integration, governed data, and metering belong in the same stack.
Read more
Both posts approach the same buyer question from the platform side.
One walks through Salesforce's acquisitions as layers of an AI operating system — Informatica for governed data, Fin for autonomous service resolution, Contentful for a headless content layer, m3ter for billing metered agent interactions — and argues that systems which only reply to text are not agents; the ones that matter start workflows, update systems, trigger transactions, and enforce policy as they act.
A sponsor selling the orchestration or enforcement layer gets an enterprise-architecture section and a 2026 roadmap hook from this framing.
The posts are written from inside a Salesforce and MuleSoft practice.
Both stay within one vendor's stack, and the second closes by inviting direct contact for the practice's services, so the argument reads as a platform position rather than a neutral buyer's view.
They lack the concrete case an article needs about the damaged order where chat approves a refund, voice declines it, and text promises a replacement.
Ask for that case; a sponsor article would otherwise inherit the Salesforce-specific framing unless rewritten.
Choose Votek when the deliverable is a comparative market scan rather than a single argument.
Read more
The agent-readiness scoring covers 22 software companies across five named dimensions — API openness, agent-native features, data portability, real-time access, and ecosystem flexibility — and reports scores including Shopify 8.6, Stripe 8.4, Salesforce 8.4, Oracle 7.2, and SAP 6.6, with a note that both Salesforce and SAP are implemented at Customertimes.
A sponsor article that places agentic customer service in a ranked landscape, or gives buyers a repeatable vendor-evaluation frame, gets more from this than from one case study.
The second post adds the professional-services angle: IBM Consulting's move toward 'service as a software', $3.5bn in savings across 70 corporate functions, and the cited finding that 95% of generative AI projects failed to deliver financial returns.
Method transparency is the weak point.
The scores are a self-published exercise: the post names five dimensions but shows neither the rubric nor the evidence behind each number.
A sponsor article would have to defend or drop the scores.
The content also sits one level above an article about routing one refund decision across chat, voice, and text: it compares platforms and discusses consulting business models, not the decision itself.
Ask for the customer-service case if that is central to the brief.
Choose Barwal when the need is a practical checklist for a mid-market operations reader.
Read more
The automation post sets out five processes worth automating first — lead qualification and routing, tier-1 support and order inquiries, cross-system data synchronisation, document verification and client onboarding, and real-time operational monitoring — each with a short before-and-after description.
The second post applies the same approach to eCommerce performance, naming heavy assets and script creep, slow checkout database queries, and slow server rendering, with fixes for each.
It notes that moving from 3.8 seconds to under 1.2 seconds can change conversion.
A sponsor wanting a how-to piece, a numbered framework, or a storefront-performance angle gets a ready structure.
The posts stay at the level of categories and standard fixes.
The support post names tier-1 order inquiries as a candidate for automation but says only that 'a properly designed AI agent can securely retrieve information from approved systems'.
It never shows a decision failing, a policy conflict, or two channels disagreeing — details an article about a refund decision across channels needs.
The performance post is about storefront speed, unrelated to a customer-service sponsor.
Both use heavy hashtag and keyword padding, so a sponsor would be buying the structure rather than the prose.
This profile draws on three written LinkedIn posts by Juan Jaysingh: one on where AI agents struggle in complex customer service, one on Salesforce and Anthropic putting the CRM inside Claude, and one on decision orchestration and headless customer experience.
One post summarises a bylined article for CRM Magazine and another records a quoted answer given to CX Today; the full articles behind them are not included here, and the posts carry no publication dates, so positions are reported as current rather than as a dated sequence.
The posts do not cover video, audio or live formats, and no audience composition, engagement data, rate card or sponsorship terms were available.
The similar-creator comparisons rest on posts by three other writers, dated November 2025 to September 2026.
Gaps in the available information are reported where they change a reader's decision rather than smoothed over.
