The Knowledge Layer Your Social Media Team Didn’t Know It Needed
If you’ve ever lost a brand guideline document because it was buried in three different Slack threads, a Notion page, and a Google Doc that nobody can find, you already understand the problem Pulse is trying to solve. Social media operators and creator teams are notorious for living inside chat apps and collaborative docs – Slack for real-time decisions, Notion for strategy, Trello for content calendars, Google Drive for asset storage. The result? Every AI assistant you talk to starts the conversation from scratch. It doesn’t know that you decided to sunset the “XYZ campaign” last Thursday. It doesn’t know that the client approved a new tone-of-voice rule only in a quick call note. It doesn’t know that the lesson from the last viral post was “shorter captions win on TikTok after 9 PM.” And because most “company knowledge” tools are either surveillance boxes or hallucination factories, you learn to distrust them quickly.
Pulse, a product launched by solo builder Apoorv Jain on Product Hunt, attempts to close that gap with a permission-aware, fully cited memory layer that lives inside the AI tools you already use – Claude, Cursor, and ChatGPT – via the Model Context Protocol (MCP). It is not a social media scheduler, not a publishing tool, and not an analytics dashboard. It is something rarer: a structured, auditable archive of decisions, commitments, and lessons that outlive the people who made them. For any creator team that has ever wasted hours re-litigating a strategy pivot because nobody could remember who said what and why, Pulse is worth a serious look. But it also arrives with sharp edges that social-media-first users need to see before they hand it the keys to their brand brain.
What Problem Pulse Actually Solves
The product’s core thesis is stated clearly by the maker: “Every AI tool I used at work had the same flaw: it didn’t actually know my company. Every session started from zero. Every answer sounded confident and gave me nothing to verify. And most ‘company knowledge’ tools quietly turned into surveillance.” That line lands hard if you’ve ever tried to use a chatbot to summarize your brand’s content strategy, only to get a generic answer that could apply to any DTC brand. Pulse treats decisions, commitments, and lessons as first-class objects, not strings buried in a Slack thread. It captures them, links them to who made the decision and why, and they stay accessible even after the person leaves the team.
For a social media manager, think about the number of times you’ve had to hunt down an old comment thread to remember why you pivoted from Reels to static carousels, or why you stopped using a certain hashtag set. Pulse formalizes that institutional memory. Every answer it returns includes a citation on every line. If Pulse cannot back a claim with a source, it says so instead of guessing – a behavior that is refreshingly rare in the current ecosystem of large language models.
The permission model is the architectural heart of the product. Instead of building a giant repository of everything and then trying to gate access after the fact, Pulse retrieves only what the user already has access to in the source systems (Slack, docs, meetings). “Pulse filters before synthesis: your permission scope is rebuilt on every query, so only documents you could open yourself ever reach the model,” Jain explained in a comment thread on the launch page. That means two people asking the same question get different answers, scoped to what each can see. This is not a bug – it’s correct behavior for a shared brain that respects existing permissions. In a creator team where a junior social coordinator should not see pricing strategy docs, but the head of content should, Pulse’s design is a genuine upgrade over tools that dump everything into a single vector store and hope for the best.
Under the hood, Pulse is a full platform: a proactive Home feed, a cited Ask feature with deep research, a decision graph, expert finder, agents with human approval, real-data outputs, and voice control. The decision graph, in particular, caught my attention. In response to a question about whether old decisions retire when reversed, Jain described a nightly job that scans for contradictions. When a new decision contradicts an old one, Pulse shows both views, marks which one is current, and explains the ranking: recency, authority, and specificity, in that order. “The reversal itself becomes useful history too,” he wrote. That’s the kind of nuanced handling that most wiki-like tools never attempt.
How It Differs from Existing Options
Notion AI, Guru, and the Knowledge-Base Graveyard
Most creator teams I know start with Notion. It’s flexible, it’s relatively cheap, and it can double as a CMS. Then they bolt on Notion AI for Q&A. The problem is that Notion AI has no native understanding of permissions beyond page-level sharing – it will happily answer questions about a page you technically cannot edit, as long as you have read access. And it has zero concept of a “decision” as a discrete entity. A pitch deck from last quarter and a Slack decision from yesterday live in completely different surfaces. Pulse’s decision graph and contradiction detection are absent from Notion AI entirely.
Then there’s Guru, which positions itself as a company wiki with verified cards and AI search. Guru does have a permission model, but it’s built around card-level access, not real-time permission rebuilding from source systems. Pulse’s approach – where your access is re-evaluated on every query by checking what you can actually open in the source – is more robust for teams that use multiple tools. Guru also lacks the agent-with-human-approval workflow that Pulse offers. If you want an AI to draft a policy change and then wait for a human sign-off before storing it as a decision, Guru doesn’t have that.
The MCP Bet: Why It Matters for Power Users
Pulse lives inside Claude, Cursor, and ChatGPT over the Model Context Protocol. That means you don’t open a separate Pulse dashboard to ask questions – you ask your existing AI assistant, and it pulls from Pulse’s knowledge layer. This is both a strength and a friction point. For creators who spend their days in Cursor writing code for automations or in ChatGPT drafting captions, the integration feels seamless. For the majority of social media managers who still use the web interface of ChatGPT or the apps, you need to be using MCP – which is still not a mainstream workflow. I’d wager fewer than 10% of the content operators reading this have set up MCP at all. Pulse’s value is therefore gated by the user’s willingness to adopt a somewhat technical infrastructure. The team provides a free demo at pulsehq.tech/onboarding, but the onboarding flow will need to be frictionless if it wants to cross the chasm from developer-friendly to social-media-team-ready.
Contrast with Internal Search Tools (Slack, Google Drive AI)
Slack itself has been rolling out AI features – summarization, Q&A, channel search. But Slack’s AI does not create a durable decision graph. Once a thread is archived, finding the reason a decision was made often requires reading 50 messages. Pulse captures the outcome and links it back to the source, so you get the “why” without the scroll. Google Workspace’s AI sidebar is similarly limited to the document you’re in. Pulse is the only tool I’ve seen that treats the composite of Slack, docs, and meetings as a single semantic graph that respects source permissions.
What Creators and Social Media Teams Can Borrow From It
Decision Tracking for Content Strategy Pivots
The most expensive thing a content team produces is not a video – it’s the institutional knowledge of why that video worked. When you change your posting schedule from daily to three times a week because analytics showed diminishing returns, that decision should be a first-class object that future you (or a new hire) can retrieve. Pulse’s decision graph lets you tag who made the call, what evidence they used (e.g., a spreadsheet of engagement rates), and when it should be re-evaluated. I’ve personally spent days onboarding to a new social team by reading old Slack threads and Notion pages. With a tool like Pulse, that process could be cut to hours.
Brand Guideline Enforcement via Citation
Every sentence Pulse outputs is linked to a source file. If an AI assistant tells you “our brand voice is conversational and witty,” Pulse will point you to the specific doc or Slack message where that rule was recorded. This is transformative for teams that have struggled to get AI to stop hallucinating brand guidelines. Imagine asking ChatGPT to “write a LinkedIn post about our new product launch” and having it cite the exact tone-of-voice doc, the approved product description, and the previous example that worked best. Pulse’s citation model makes that possible – provided the source materials are properly ingested.
Agents with Human Approval for Automated Publishing
Pulse includes agents that can take actions, but only with human approval. For a social media operator, this could mean an agent that drafts a content calendar, pulls past performance data, and suggests a posting time – then waits for your sign-off before writing it to the decision graph. While Pulse is not a scheduler itself (no direct integration with Buffer or Hootsuite), the approval workflow concept is directly applicable. If Pulse ever adds an API to push decisions to a calendar tool, it would become a powerful bridge between strategy capture and execution.
Where the Math Breaks
Not a Social Media Management Tool – and That’s Fine, but Limited
Pulse does not schedule posts, does not track engagement, does not pull analytics from platforms. It is a knowledge layer, not a command center. If you are a solo creator looking for a one-stop shop to plan, publish, and analyze content, Pulse will not replace your current stack. It’s best suited for teams of three or more who already have a publishing tool (like Buffer, Later, or Metricool) and need a back-end brain. For a one-person operation, the overhead of capturing decisions may outweigh the benefit – you already know what you decided because you were the only one in the room.
Permission Model: Strong for Internal, Fragile for Agencies
The permission-rebuilding approach is smart for internal teams where access control is well-defined. But for social media agencies that serve multiple brands, the model could break quickly. If a freelancer collaborates on a campaign but should not see the client’s financial data, the source systems (Slack, Google Drive) need to enforce that access correctly. Pulse inherits the flaws of its source permissions. If your Slack workspace gives a contractor access to a channel with sensitive information by accident, Pulse will surface that information. It’s not a security panacea – it’s only as good as the permissions you’ve already configured. The maker acknowledges this by asking for honest takes on where it would break for teams.
The Pricing Gap
Pulse’s pricing is not disclosed in the launch material. For indie creators and small teams, cost is a major factor. If Pulse charges per seat like most enterprise knowledge tools, a three-person content team may balk. If it offers a free tier or flat-rate pricing, it could compete with Notion AI. But the silence on pricing means I cannot assess whether it’s affordable for the audience this essay targets. I’ll be watching for that number.
Overhead of Decision Recording
The decision graph is powerful, but it requires active work from the team to record decisions, commitments, and lessons. If Pulse cannot passively consume Slack conversations and extract decisions automatically, the burden falls on humans to log every pivot. In my experience, most teams are terrible at documentation, especially in fast-moving social media where “we’ll document it later” never happens. Pulse’s proactive Home and expert finder suggest some degree of automation, but the specifics are not clear. The maker’s description of a nightly job that scans for contradictions implies Pulse does some passive detection, but the initial capture may still require intentional tagging. Teams with low documentation discipline will likely see a graveyard of unlabeled content.
Why TikTok Creators Should Care More Than LinkedIn Ones
TikTok content moves fast. Trends shift weekly, platform algorithms change overnight, and the difference between a winner and a dud is often a small strategic decision (e.g., “use this sound, not that one”). Institutional memory is brittle in a high-velocity environment. Pulse’s decision graph could help a TikTok team track which sound strategies worked, which hooks outperformed, and why a particular style was retired – all linked to actual conversations and data. LinkedIn creators, by contrast, often operate in a slower, more deliberate cadence. The value of a cited, persistent memory layer is lower when decisions are made monthly, not daily. If you run a TikTok-heavy operation, Pulse’s ROI might justify the complexity.
Where the Math Breaks: Permission Model in Agencies
A common scenario: an agency manages three brands. Each brand has its own Google Drive folder and Slack channels. A social media manager who works on Brand A should not see Brand B’s unreleased campaign. Pulse’s permission model will respect that if the agency has set up separate workspaces or proper folder-level permissions. But many agencies grant contractors broad access to “Client X – Full Access” for convenience. In that case, Pulse will surface more than intended. The maker’s answer about “two people asking the same thing get different answers scoped to what each can see” sounds ideal, but it assumes the source permissions are correct. In messy agency setups, Pulse could leak context across clients through the AI assistant’s answers, even if citations are filtered. The risk is real and should be tested with dummy data before rolling it out.
What I’d Watch / Test Next
I’m going to sign up for the free demo this week with a small social media team I consult for. My test plan: feed Pulse a few real decisions from our last campaign – why we switched from Reels to carousels, which caption templates performed best, the exact date we sunsetted a brand persona. Then I’ll ask Claude (connected via MCP) the same question two weeks later and see if the answer retains the context and cites properly. I’ll also simulate the permission test by having a junior member ask about a decision they shouldn’t have access to, to confirm Pulse respects the source permissions correctly.
If that works, the next step is to evaluate the friction of recording decisions. Does Pulse integrate with Slack natively to capture decisions from threads, or do we need to manually type every lesson? The launch page mentions Slack integration (via source access), but the exact capture mechanism is under-documented. I’ll test whether Pulse can automatically extract a decision from a Slack message where we said “Let’s go with the vertical format only” and tag it as a decision.
For creators and social media operators who are tired of AI tools that sound confident but have no memory, Pulse is worth a serious evaluation – if you are willing to invest in the setup. If you work solo or have a tiny team that already documents everything in a single Notion page, skip it. But if you manage a team of three or more, with decisions flying across Slack, docs, and meetings, Pulse could be the knowledge layer that stops the re-litigation cycle. I’ll report back after my tests.
