Joseph Paul is a MongoDB Customer Success Manager who works on agentic AI and agent memory, including building agents for MongoDB’s go-to-market teams.
Joseph Paul is a MongoDB Customer Success Manager whose posts connect agentic AI ideas with practical work inside a go-to-market organization.
Their clearest example is an eight-layer framework for agent memory, from short-lived working memory to shared memory across agents.
They also write about hands-on technical training, customer success, and a career path from enterprise sales at NI into software and AI work.
Follow them for accessible explanations of agentic memory and examples of how AI tools can support business teams.
Paul makes a complex topic easier to discuss by separating agent memory into eight jobs and explaining what each contributes.
Episodic memory, for example, records what happened and why, while semantic memory distills more stable knowledge from those records.
The distinction shows why a context window alone is not persistent memory.
The posts also connect concepts to practice: Paul says agents built for MongoDB’s go-to-market organization saved hundreds or thousands of hours, and describes hands-on skill badges as a way to reinforce application-design and performance practices.
Together, the posts show a mix of technical explanation, team enablement, and sales experience.
Joseph Paul writes about agentic memory as a layered engineering problem and about making that work measurable for a business.
Read more
Two priorities run through the posts: build agents whose value shows up as saved time, and train technical teams on the actual product rather than relying on documentation alone. An earlier commercial career shows the same habit of justifying technology through workflow analysis and return on investment.
Agent memory is eight problems, not one
Joseph Paul argues that agentic memory is treated as a single feature when it is really eight interacting layers: working, episodic, semantic, procedural, entity, taxonomic, long-term, and shared.
Read more
Treat the memory engine as one unit, they say, and the system will forget, confuse, or hallucinate context.
Each layer has a specific job.
Working memory is the scratchpad for the current conversation and tool results; it dies when the session ends.
Episodic memory records what happened, why, and the outcome.
Semantic memory holds distilled facts.
Procedural memory captures learned workflows.
Entity memory tracks people, companies, and their relationships.
Taxonomic memory organizes how knowledge is classified.
Long-term memory handles persistence, compression, and decay.
Shared memory lets one agent's knowledge reach another.
Their complaint is that most frameworks ship working memory and maybe a basic version of semantic memory, leaving the other six for developers to build—which is why persistent memory remains hard.
The entity layer is described as the agent's CRM: a record that Sarah is the VP of Engineering, reports to Mike, and prefers asynchronous communication.
The post follows an earlier one on persistent memory and promises a later piece mapping MongoDB products to each layer, building the argument as a series from problem to product.
Joseph Paul's test for agent work is operational: agents should remove real work from a go-to-market organization.
Read more
They say that, over two years at MongoDB, they specialized in agents and agentic memory and built agents that saved hundreds or thousands of hours across the GTM org.
A nomination followed to help shape the AI tools the organization uses for productivity.
This is a practitioner's priority, not a research agenda.
It sits alongside the technical framework: the layers explain the engineering; the saved hours explain why anyone should care.
The posts describe internal scale: agents credited with saving hundreds or thousands of hours across a sales and customer-success organization, alongside a role shaping which AI tools that organization adopts.
The AI interest was declared before the job existed.
Joseph Paul's first-day post says they were drawn to MongoDB by the chance to work with technology in a passionate space.
Their two-year post reports that the earlier goal of working with AI and moving outside their comfort zone was met by moving from sales at a test-and-measurement company into customer success at a software company.
Joseph Paul argues that intentional training is the best way to make applications scale efficiently, framing hands-on instruction as a way to avoid costly design and performance mistakes rather than a box to tick.
Read more
They recommend MongoDB Skill Badges to reinforce best practices, pointing to MongoDB's own Customer Success organization, which spends time earning badges. The post ends with an offer to set up a Skill Badge Day for any team that wants one.
Joseph Paul offers to help teams arrange a Skill Badge Day and promises to share which badges were earned and which teams or use cases each one fits.
Sell technology by mapping a customer's workflow to a number
Before the AI work, Joseph Paul's software-selling method was to analyze a customer's workflow, quantify bottlenecks, and calculate return on investment so a manager could justify the purchase.
Read more
The other half of that method was building relationships: cultivate champions among midlevel managers and influential engineers, then give them a stage.
They hosted yearly conferences where those champions could present their collaborations and lessons learned to a global peer group.
The same instinct—turning a technical claim into a figure a leader can approve—later shows up in descriptions of agent work through saved hours.
At NI, Joseph Paul says the strategy covered the company's largest Power & Performance customer, drove about $18 million in revenue over 2.5 years, and grew enterprise software annual recurring revenue to about $370,000. The framework focused on selling software through workflow analysis and ROI.
The audience changed. The earlier ROI method targeted buyers and their leadership; the newer agentic memory material is written for developers and technical teams, though it still points toward product value.
The framework is presented inside one company's products
Joseph Paul's technical and educational writing is anchored in MongoDB.
Read more
The eight-layer memory post promises a follow-up mapping MongoDB products to each layer; the training post recommends MongoDB Skill Badges; the two-year reflection mentions MongoDB announcements at a .local event. Readers should treat the product context as part of the position, not a detail.
The same pattern appears in all three places: a memory framework that ends with a MongoDB product mapping, a badge recommendation tied to MongoDB's Customer Success team, and a career post built around MongoDB milestones.
Explains what agentic memory includes and why persistent memory is treated as eight interacting problems rather than one feature.
Read more
Start here for the clearest statement of Joseph Paul's technical framework.
Joseph Paul lays out an eight-layer framework for agent memory: working, episodic, semantic, procedural, entity, taxonomic, long-term, and shared.
Each layer gets a plain-language explanation.
Working memory is the scratchpad that dies with the session.
Episodic memory records what happened, why, and the outcome.
Semantic memory distills stable facts.
Procedural memory captures learned workflows.
Entity memory tracks people, companies, and relationships.
Taxonomic memory organizes concepts.
Long-term memory handles persistence, compression, and decay.
Shared memory lets multiple agents access the same knowledge.
They argue most frameworks today provide only working memory and maybe basic semantic memory, leaving the other six to the developer.
The post closes by promising a later mapping of MongoDB products to each layer.
For brands planning a campaign
For a developer tools or AI infrastructure company, this framework could anchor a technical explainer series or webinar: one segment per layer, with product mapping as the payoff. The eight-layer model gives a sponsor a neutral educational spine rather than a product pitch.
Explains what Joseph Paul has built and how their interest in agentic memory connects to work inside a GTM organization.
Read more
Newcomers can read it for the practitioner context behind the framework.
Joseph Paul marks two years at MongoDB and describes a deliberate pivot: after five years at NI and a layoff, they wanted to work with AI and move outside their comfort zone.
They say the last two years were spent specializing in agents and agentic memory and building agents that saved hundreds or thousands of hours across a go-to-market organization.
They also mention a nomination to help shape future AI tools for the organization and contrast prior sales at a test-and-measurement company with customer success at a software company.
Personal context includes expecting a first child and excitement about MongoDB announcements.
For brands planning a campaign
Useful for campaigns about internal enablement, customer success operations, or applied AI in sales. The post shows operational agent work, not just conceptual writing, making Joseph Paul a fit for practitioner-facing case content.
Shows how Joseph Paul approaches learning and enablement: through concrete training, badges, and team practice.
Read more
Start here to see the teaching style behind the more technical memory post.
Joseph Paul argues that intentional training is the best way to make applications scale efficiently and recommends MongoDB Skill Badges as a practical, hands-on way to reinforce best practices and avoid costly design or performance mistakes. They note that MongoDB's Customer Success organization spent time earning badges and offer to help teams set up a Skill Badge Day.
For brands planning a campaign
A certification, training, or developer education brand could adapt this into a co-hosted workshop, badge day, or enablement campaign. The post already includes a call to action they are willing to run.
Explains when and how Joseph Paul entered MongoDB and customer success.
Read more
It offers a short starting point on timeline and motivation, especially before the two-year reflection.
Joseph Paul announces their first day at MongoDB as a Customer Success Manager and thanks people in their network who helped through referrals, encouragement, and interview preparation, naming Kody Jones, Stephen Reda, and Tom G. They express excitement about working with technology in an area they are passionate about.
For brands planning a campaign
Limited direct campaign application, but the post establishes the customer success role and the network-driven path into MongoDB. It could support a story about career transitions into developer-focused customer success.
Explains what Joseph Paul did before MongoDB and why the later move into customer success marked a change of lane.
Read more
Newcomers can read it for the sales and account background that shapes their later GTM agent work.
Joseph Paul announces the end of their role at NI after layoffs and lists target roles: account management, account executive, enterprise software sales, business development, sales strategy, SDR management, sales enablement, GTM strategy, and cross-functional project management. The post also records their accomplishments: growing NI's largest Power & Performance customer, leading a cross-functional global team, driving about $18 million in revenue over 2.5 years, growing enterprise software annual recurring revenue to about $370,000, building a framework to sell software through workflow analysis and ROI, and cultivating champions across management and engineering.
For brands planning a campaign
Best suited to sales enablement or GTM strategy campaigns rather than agentic memory content. It offers evidence of enterprise sales experience and a framework for translating technical value into ROI.
Building agents that save go-to-market teams time1
Skill badges and intentional training as adoption levers1
Agentic memory as an eight-layer problem
Joseph Paul's central technical argument is that agentic memory is not one feature but eight interacting layers: working, episodic, semantic, procedural, entity, taxonomic, long-term, and shared.
Read more
They warn that treating a memory engine as one unit causes agents to forget, confuse, or hallucinate context.
Most agent frameworks, they note, provide working memory and maybe basic semantic memory, while leaving the other six layers to developers.
The post ends by promising a follow-up mapping MongoDB products to each layer.
Defines each layer in plain terms, then groups them into a reusable framework. Uses a series format: a previous post on persistent memory, this post on the eight layers, and a next post on product mapping.
For brands planning a campaign
A developer tools or AI infrastructure company can build a technical explainer series or webinar in which Joseph Paul walks through the eight layers and helps audiences map their own stacks to them. The framework gives the campaign a ready-made episode structure and a clear distinction between working and persistent memory.
In a two-year MongoDB reflection, Joseph Paul says they have spent two years building expertise in agents and agentic memory, and developing agents that saved hundreds or thousands of hours across the GTM organization.
Read more
They were nominated to help shape future AI tools the organization uses to maximize productivity.
Connects technical agent work to operational outcomes: hours saved and productivity gains across a sales and customer-success organization.
For brands planning a campaign
For an AI infrastructure or productivity platform, Joseph Paul can discuss internal agent deployment from a practitioner's perspective: what was built, which workflows it touched, and how a GTM team measured time saved. This suits case studies, webinars, or field guides aimed at operators as well as engineers.
Skill badges and intentional training as adoption levers
Joseph Paul argues that intentional training helps applications scale efficiently.
Read more
They recommend MongoDB Skill Badges as a hands-on way to reinforce best practices and avoid costly design or performance mistakes. They cite MongoDB's Customer Success organization earning badges and offer to help set up a Skill Badge Day.
Connects training to risk reduction and performance, then gives teams a direct call to action. Offers to share which badges they earned and which teams or use cases suit them.
For brands planning a campaign
A training, certification, or developer platform can book Joseph Paul for a webinar or workshop on hands-on enablement, badge programs, or team skill-building. The Skill Badge Day offer gives the campaign a concrete, repeatable format rather than a generic thought-leadership slot.
Earlier posts show Joseph Paul's background in account management, enterprise software sales, business development, sales enablement, and GTM strategy.
Read more
At NI, they say they led global Power & Performance and Enterprise SW strategy, drove about $18 million of revenue in 2.5 years, grew annual recurring revenue to about $370,000, built a framework for selling software by analyzing customer workflows and quantifying bottlenecks and ROI, and cultivated champions across midlevel management and engineering.
Uses quantified outcomes and a repeatable sales framework: map workflows, identify bottlenecks, calculate ROI, and justify purchases to leadership.
For brands planning a campaign
For B2B software or sales-tech brands, Joseph Paul can create content on translating technical value into business cases, executive champions, and ROI-led selling. This complements their newer agentic AI content and can help a campaign reach sales and customer-success audiences.
Joseph Paul describes joining NI straight out of college, working there for five years, and being laid off.
Read more
After a job search, they joined MongoDB as a Customer Success Manager because they wanted to work with AI and push beyond their comfort zone. Two years later, they describe moving from sales at a test-and-measurement company into customer success at a fast-paced software company and building AI expertise on the job.
Personal career reflections that name a goal, describe the transition, and report progress. The posts often thank specific people and networks.
Not a technical topic; the career posts do not teach agentic memory or AI engineering.
For brands planning a campaign
Employer-brand, career-development, or community-focused campaigns could draw on Joseph Paul's transition stories. Their career reflections also provide context for the learner's perspective in their technical content.
Joseph Paul's posts place the subject matter inside MongoDB: they joined as a Customer Success Manager, recommend MongoDB Skill Badges, plan to map MongoDB products to memory layers, and mention MongoDB announcements at a .local event.
Read more
The agentic memory framework is the main technical idea, presented in a MongoDB product context.
MongoDB-anchored educational and workplace posts that use company products, events, and training as their reference frame.
For brands planning a campaign
Brands in or around the MongoDB ecosystem, such as complementary AI infrastructure or enablement tools, can evaluate Joseph Paul for co-educational content. Direct competitors should treat the MongoDB context as a material fit question rather than assume neutral technology commentary.
Joseph Paul's posts describe employment relationships.
Read more
Paul works at MongoDB as a Customer Success Manager and has built agentic AI tools for MongoDB's go-to-market teams; before that Paul spent about five years at NI (formerly National Instruments), which Paul left after a layoff. Joseph Paul's posts do not mention sponsorships, affiliate arrangements, paid partnerships, or client engagements, and no third-party brand is presented as a commercial partner.
MongoDB
Employment
Paul works at MongoDB as a Customer Success Manager.
Read more
They announced their first day in the role, later marked two years at the company, and say they have spent that time building agents and developing expertise in agentic memory for MongoDB's go-to-market organization. They also promote MongoDB Skill Badges, including an internal skill-badge day for the Customer Success org, and offer to help teams set up their own.
Paul joined NI straight out of college and worked there about five years before being laid off. In the role described in the post, they managed NI's largest Power & Performance customer, led enterprise software strategy globally, and reported roughly $18MM in revenue over 2.5 years.
Start with Joseph Paul’s eight-layer agentic memory post, then read their two-year MongoDB reflection on time saved inside a go-to-market organization.
Read more
The Skill Badges post covers training and enablement, while the earlier NI job-search post supplies the commercial method behind the later AI work.
The memory piece refers to a prior post on persistent memory and promises a later mapping of MongoDB products to each layer; ask for that mapping to assess product fit.
For background on ROI-led selling, read the NI post.
Both explain AI memory and context in practical terms for business and technical teams.
Read more
Terry Miles discusses short- and long-term memory in context engineering and how procurement teams can start with AI questions that use their data. Joseph Paul offers a layered memory framework and a training path for teams adopting AI.
Terry Miles writes from a procurement and instructional-design perspective, using procurement data questions and SLMs as the entry point. Joseph Paul writes from a MongoDB customer-success and go-to-market perspective, using an eight-layer memory taxonomy and internal agent deployments as the reference frame.
Both write about how enterprise software decisions affect buyers and users.
Read more
Joseph Paul’s earlier work explains ROI-led selling and workflow analysis; Adam Mansfield explains what acquisitions, pricing, and licensing changes mean for software customers and procurement teams.
Adam Mansfield focuses on procurement leverage and vendor moves in the Salesforce ecosystem. Joseph Paul focuses on agentic memory, internal AI agents, and training inside a customer-success organization, with enterprise sales as background rather than the current subject.
Both write about why AI agents fail when context is fragmented, and translate memory or context-layer concepts into explanations a technical audience can use.
Read more
Conor Brennan-Burke describes agents as blind mice touching different parts of an elephant; Joseph Paul describes memory as eight interacting layers.
Conor Brennan-Burke writes as a founder building a context layer at Hyperspell, critiquing connector-based agent stacks and staking out a product position. Joseph Paul writes as a MongoDB customer-success practitioner, offering a layered memory framework and promising a mapping to MongoDB products.
For a developer-tools or AI-infrastructure campaign, Joseph Paul’s posts position them as a technical explainer on agentic memory, not yet as a demonstrated reach channel.
Read more
The only measured audience figure is 1,530 LinkedIn followers as of 23 October 2026; the posts don’t describe their composition or show how many developers, data engineers, or AI practitioners see each post.
Follower totals don’t reveal employer, occupation, or seniority.
The writing is aimed at people building or buying agent systems.
The August 2026 eight-layer breakdown of agentic memory says, “Most agent frameworks today give you layer 1 and maybe a basic version of layer 3… The other six are left as an exercise for the developer.”
The December 2025 post promotes MongoDB skill badges to teams seeking to avoid design and performance mistakes.
Earlier posts—a June 2024 layoff announcement and a September 2024 first-day-at-MongoDB note—address a personal and professional network, offering little evidence of topical audience fit.
The September 2026 two-year reflection describes agents built for MongoDB’s own go-to-market teams, demonstrating hands-on production experience rather than revealing who reads the public posts.
Before committing budget, ask Paul for per-post LinkedIn analytics on the agentic memory series—impressions, reactions, comments and, where visible, commenters’ job titles and roles—along with any audience survey or newsletter list reporting developer seniority.
Request figures for a comparable technical explainer, since topic, depth, and format shape results.
LinkedIn followers
1,530
LinkedIn profile (follower count shown on the account page) · 2026-10-23
Joseph Paul's public posts cover their customer-success work at MongoDB, a MongoDB Skill Badge training day, an eight-layer framework for agentic memory in AI systems, and two career milestones.
Read more
They do not describe a sponsorship offer, rate card, media kit, partnership program, paid brand collaboration, sponsored format, or official brand inquiry channel.
Ask directly about availability and terms before budgeting a paid placement.
Their content is tied to their employer's product area, so any commercial arrangement would sit alongside their MongoDB role rather than operate as a separate independent media business.
A developer-tools or AI-infrastructure company wants educational content on agentic memory and proposes a technical explainer series or webinar built on Joseph Paul's eight-layer framework.
Read more
Verdict: good fit, with conditions.
The framework is already a usable episode spine.
Joseph Paul names and defines eight layers — working, episodic, semantic, procedural, entity, taxonomic, long-term, and shared — and gives each a plain-language job: scratchpad, journal, world understanding, learned workflow, the agent's CRM, how knowledge is organized, the persistence and consolidation layer, and memory shared across agents.
The payoff is concrete for a builder audience: treat memory as one unit and the system will forget, confuse, or hallucinate context; most frameworks provide only working memory and maybe basic semantic memory, leaving the other six to the developer.
That is a ready-made structure for a four-to-six segment series, and the analogies travel well to slides and video.
They also speak from work rather than commentary.
They say they spent two years specializing in agents and agentic memory, built agents that saved hundreds or thousands of hours across a go-to-market organization, and were nominated to help shape the organization's future AI tooling.
For a sponsor that wants practitioner framing — what was built, which workflows it touched, and how the team judged the benefit — that is useful positioning.
The posts include one entry in the series, dated 25 August 2026; the earlier persistent-memory post and the promised MongoDB product-mapping post are referenced but not published here.
Ask for those materials before scoping the series.
The content is anchored in MongoDB products and events, so the sponsor must accept an ecosystem-adjacent framing; a direct MongoDB competitor should treat this as a material fit question, not neutral technology commentary.
Contract and employer permission are unconfirmed, and no sponsorship format or inquiry route appears in the posts.
Ask about approval, terms, and disclosure before committing.
The only measured audience figure is 1,530 LinkedIn followers on 23 October 2026; ask for audience composition and per-post metrics before treating this as a reach buy.
When to choose a specialist
A consumer or enterprise brand wants a paid, high-reach sponsored placement — a rate-card booking of a LinkedIn post or product video with guaranteed impressions and a documented media kit.
Read more
Verdict: poor fit based on the posts here.
No post states a sponsorship offer, rate card, media kit, pricing, or inquiry route.
Reach cannot be promised either.
The single measured audience figure is 1,530 LinkedIn followers, which says nothing about how many developers, data engineers, or AI practitioners see a post, or their seniority.
No impressions, reactions, comments, or video performance data are published; topic, depth, and format all shape results.
The content also sits inside an employer context.
Joseph Paul works at MongoDB, recommends MongoDB skill badges, and plans to map MongoDB products to the memory layers.
Any commercial arrangement would sit alongside that role, making employer approval, disclosure, and timing open questions.
A paid placement that needs hard product calls to action and measurable conversion from a single sponsored post would be better served by a creator with documented sponsorship formats and verified audience analytics.
Before reconsidering this option, ask for sponsorship terms and post-level performance data.
A worked campaign example
Worked angle: an eight-layer memory clinic for a developer-tools or AI-infrastructure sponsor.
Format: one 45-minute live webinar plus a three-post LinkedIn companion series, hosted by Joseph Paul, with a named engineer from the sponsor joining the final segment.
Segment 1, about 15 minutes — why one memory engine fails.
Open with the problem they state directly: if you treat memory as a single unit, your agent will forget, confuse, or hallucinate context.
Then walk through layers 1 through 4 using the post's own comparisons: working memory as the scratchpad that dies with the session, episodic memory as the journal that records what happened, why, and the outcome, semantic memory as distilled facts, and procedural memory as learned workflows so agents stop re-learning the same task each session.
Segment 2, about 15 minutes — layers 5 through 8 and where systems break.
Cover entity memory as the agent's CRM for people, companies, and relationships; taxonomic memory as how an agent organizes what it knows; long-term memory as persistence, compression, decay, and promotion or demotion by relevance; and shared memory as the point where single-agent architectures break down and multi-agent systems need a shared backend.
Shared memory is the natural place to raise the security question they flag, and a good hook for the sponsor's own product discussion.
Segment 3, about 10 minutes — mapping exercise.
Attendees map their own stack against the eight layers on a one-page worksheet and mark which layers they have, which they assume they have, and which they have never addressed.
The sponsor's engineer then shows one mapping live, in plain terms, without a feature tour.
Close with 5 to 10 minutes of questions, collected before the session so the host can group them by layer.
Companion series: three LinkedIn posts over three weeks — one on layers 1 to 4, one on layers 5 to 8 with the shared-memory security note, and one carrying the worksheet and the recording. Each post should keep their teaching style: define the layer, give the concrete failure it prevents, then point to the worksheet.
What the sponsor supplies: the brief, the product map the engineer will show, registration and landing page, and post promotion. What they supply: the framework, the teaching, and the discussion.
Verify before committing budget: whether their employer permits a paid external engagement and how it would be disclosed; per-post impressions, reactions, and commenter roles for the 25 August 2026 memory post and, if available, the earlier persistent-memory post; whether the referenced product-mapping follow-up can be adapted for the sponsor; and whether MongoDB can be named. If the sponsor competes directly with MongoDB, expect the independent-comparison version of this angle to be unavailable and choose a different creator.
Terry Miles is the better choice when a sponsor needs a teaching-led explainer on context engineering and memory for procurement, IT, and other enterprise stakeholders rather than application developers.
Read more
The posts offer a practical sequence: start with a narrow, valuable question and run toward complex legacy data (23 July 2025), then define context engineering and short- and long-term memory in instructional-design terms, including confidentiality and procedural versus episodic memory (24 July 2025). For a webinar that helps buyers scope an AI initiative and bring IT and security allies in early, Terry Miles offers a more governance-oriented teaching frame than Joseph Paul's eight-layer build taxonomy.
The tradeoff is less developer implementation depth.
Joseph Paul's posts do not present an eight-layer memory model, agent build examples, or claimed hours saved; the memory discussion closely reads another writer's Decoding ML posts.
Sponsors wanting engineers to map working, episodic, semantic, procedural, entity, taxonomic, long-term, and shared memory to their own stack would lose the ready-made curriculum and practitioner build story.
The profile covers two posts; ask for audience-composition data and commercial terms before booking.
Conor Brennan-Burke is the better choice when a sponsor wants a founder-led, product-category argument about why agent context fails and what a context layer should do.
Read more
The 25 March 2026 post argues that most business context is unstructured and lives across Slack, email, docs, and CRM; the 30 June 2026 post uses the blind-mice parable to explain that more connectors give an agent more fragments without resolution. For technical buyers already convinced memory matters and now evaluating architectures, that failure narrative and product point of view can be sharper than a neutral eight-layer explainer.
The tradeoff is promotional narrowness and less curriculum.
Both posts advance Hyperspell's context-layer position, so the sponsor would be buying a founder narrative with a product agenda, not a vendor-neutral teaching framework.
They do not break memory into eight layers, offer a worksheet or segment structure, or show measured audience results.
Sponsors needing a modular educational series or seeking to avoid tying content to one startup's category claim should prefer Joseph Paul's framework and revise it for their product.
Five full LinkedIn posts by Joseph Paul and six comparison posts from Terry Miles, Adam Mansfield, and Conor Brennan-Burke provide context for the work.
The posts cover agentic memory, AI agents, training, customer success, and earlier enterprise sales work.
Joseph Paul’s posts do not show publication dates, so sequence is inferred from content.
The agentic memory series is represented by one post; the earlier and later installments it references are not included here.
No other channels, transcripts, books, shows, post-level analytics, sponsorship terms, or third-party analyses are included here; the only audience figure available is a LinkedIn follower count of 1,530 observed on 23 October 2026.
