Skip to content
Rishabh Salian
  • Work
  • About
  • Contact
  1. Home/
  2. Work/
  3. Follow
the invisible teammate

Capstone · MS Strategic Design & Management, Parsons · 2025–26

Follow.

AI made everyone faster. Follow makes teams smarter.

Follow is a shared, trackable memory layer that sits between your team’s AI tools — so what one person works out with their AI, the whole team can find, trust, and build on.

Project timeline

Spring 2025 → May 2026 — field research, expert interviews, and the two pivots that turned Follow from an AI-native document tool into a team-memory layer.

Field researchExpert interviewProduct momentPivot
PIVOT 1 · MAR 17doc editor → workspace ext.PIVOT 2 · LATE APRindividuals → teams-firstSPRING ’25FEB ’26MAR ’26APR ’26MAY ’26Housing Works3-person team · the originConcept v1cognitive debt · AI-native docPeer tests4 peers · early wireframeAlexandra Beckerconcept test · outside view · Mar 31WorkshopIntelligence Gap · 4 teamsAndyexpert · agentic systemsBuild sprintsMCP tools · the pipelineAtanu Sinhaexpert · 25-yr operatorv6 capstoneteams-first · this artifact

What this is not: a longitudinal deployment study. The peer tests informed a pivot, not a validation — the longitudinal pilot ahead is the move from modeled to measured.

The research

Mixed-methods, and honest about scope: a theoretical spine, plus six primary engagements — from lived experience to expert interviews.

Secondary · the theoretical spine

Wegner — transactive memory systems
Clark & Chalmers — the extended mind
Sweller — cognitive load theory
Bienefeld et al. — TMS in human–AI teams
Kosmyna et al. (MIT) — cognitive debt
Doshi & Hauser — homogenization risk

Primary · six engagements

Housing Works NYCLived experience
Intelligence Gap workshop4 teams · 12 participants
Peer concept tests4 peers · early wireframe
AndyExpert · agentic systems
Alexandra BeckerConcept test · outside perspective
Atanu SinhaExpert · 25-year operator

The problem

Work isn’t written alone anymore. Almost every decision now passes through someone’s AI chat — the constraints, the rejected options, the reasoning — and then stays locked in that one private thread.

Teams already run as a transactive memory system with each other: everyone keeps a rough map of who knows what. But they don’t have that with their AI tools. The context that actually shaped the work scatters across separate chats and disappears. Across five concept tests, most people volunteered the same feeling before seeing anything — that their AI-assisted work didn’t quite feel like theirs.

Transactive memory system

A group’s shared memory: instead of everyone knowing everything, members specialize and quietly keep a directory of who knows what — so the team retrieves knowledge by knowing whom to ask. Coined by Daniel Wegner, 1985.

The structural change · today

Same team. Same tools. The reasoning scatters.

Before Follow
WebAI₁AI₂AI₃NO SHARED MEMORYMayaAlexSamTHE MEETING
Each teammate works with their own AI in a private thread; the only place it all meets is a meeting — which forgets what it doesn’t know.

From the capstone deck — three teammates, three AIs, each working in a private thread the others can’t see.

Insights & areas of opportunity

Three findings shaped where Follow plays — and where the opportunity is largest.

01

AI is the invisible teammate.

Teams already run as transactive memory systems with each other — but not with their AI, because its contributions were never captured in a form the team could route to.

02

Cross-tool memory is structurally vacant.

Native memory inside one AI tool is solved by vendors. Cross-tool, cross-contributor memory is empty — no vendor with the surface to build it has a reason to make it cross-vendor.

03

Provenance matters where stakes are high.

The strongest signal for value comes where the cost of being wrong is asymmetric — legal review, regulated work, compliance. That’s also where pricing tolerance is highest.

“The deliverable shipped — we didn’t fail. But I could have done better work if I’d known how my teammates got where they got.”

From the Housing Works experience — and the five concept tests kept surfacing the same feeling, unprompted.

The opportunity

How might we give a team one shared, trustable memory — across every AI tool they already use?

The response

Follow.

A shared memory layer that lives between your AI tools — not inside any one of them.

With Follow
WebAI₁AI₂AI₃SHARED LAYERThe Follow Indexshared memory · provenance · contradiction · directoryMayaAlexSam
The Follow Index sits under every AI tool — each reads and writes the same memory, so the whole team shares it, with provenance and contradictions surfaced.

It sits between your AI tools rather than inside any one of them, and it’s queryable from any of them through MCP — the shared, scoped, conflict-aware memory is the product. (Tools like Glean index your documents; Follow indexes the reasoning.)

How it works

From conversation to queryable memory.

Follow · how it works
Maya · ClaudeAlex · ChatGPTSam · GeminiTHE FOLLOW INDEXversionedversionedversionedCONTENT 768dCAUSAL 512dCONTEXT · whoAI tool · via MCPanswer · Maya + Alex
references
supports
elaborates
supersedes
contradicts

Not quite RAG

RAG finds matching text. Follow finds the source — who worked it out, in which chat, and what it connects to.

Less a search box, more a directory of what your team knows and where it came from.

Illustrative animation of the live pipeline — three teammates and their AIs, five LLM roles, three tensors, five typed edges, and supersession chains that are never overwritten.

The sandbox

The shipped dashboard, replicated — live.

A small, pre-loaded workspace: Maya, Alex, and Sam, three different AI tools, one memory. Land on All items and see the whole week — 16 captured conversations, 7 uploaded files, and every fact they produced, newest first. Open a conversation and read the full transcript: the reasoning, the tool calls, the moment the memory got checked before anyone answered. Open a file and read the PRD, the usability notes, the analytics export the team actually indexed. Browse Facts with per-entry provenance, open the who-knows-what directory Follow keeps automatically, and ask it what's contested, who to ask, or what changed — answers come back attributed, with disagreements flagged instead of resolved. Then open the MCP console and watch the machine side: a live model calling Follow's actual tools — query_index, directory_query, detect_contradictions — with every call and result on the wire, straight from the workspace-platform repo's contracts.

↻ app.follow.team/w/aurora-checkout
Follow.sandbox
Aurora — checkout redesign3 teammates · 3 AI tools · one memory
memory synced · 32 entries

views

the team

MMayaClaude11
AAlexChatGPT11
SSamGemini10

this workspace

Facts captured32
Topics tracked8
Contested pairs3
Last capturetoday

All items

Every captured conversation, uploaded file, and extracted fact — one feed, newest first.

loading the workspace…

MCP console

Follow is headless by design — the shipped server exposes these tools over JSON-RPC at /mcp, to people and machines alike. Same names, schemas, and response shapes here.

A real model, running Follow's real tools. It decides which to call; every call and result crosses the wire below; the answer is grounded in what came back.

⚙ query_index⚙ directory_query⚙ detect_contradictions⚙ get_activity⚙ save_conversationsource: workspace-platform ↗

Try one of the prompts below — the model will pick its tools, and you'll see the JSON-RPC-shaped traffic in the open. Ask it to save the conversation and this session appears in Conversations and Facts, under “You”.

Ask Followlive model
Hi — I answer from this workspace's shared memory, with provenance. Ask what's contested, who to ask, or what changed.
Concept sandbox · pre-loaded sample workspace — the shipped Follow product captures real AI threads over MCPanswers run on a live model API via a server-side proxy
LIVE DEMO · Follow★ WORKING

The sandbox above is live.

This is the full dashboard replica: captured conversations with the MCP tool loop on the wire — thinking, tool calls, results — reading exactly like the shipped console; uploaded files rendered from their own markdown; and the fact index both produce, with per-entry provenance and three genuinely contested pairs Follow flags instead of resolving. All of it runs on a pre-loaded sample workspace; Ask Follow answers from that memory through a real model API. The MCP console goes further: a real tool-calling loop over Follow's actual MCP tool contracts — same names, schemas, and response shapes as the shipped headless server (12 tools over JSON-RPC), executed on the sandbox. Saving a conversation there still writes a real entry into every view above.

Open the sandbox full-screen

What it accomplishes

What a team gets once its AI work shares one memory.

01

Shared AI memory

One team memory across collaborators — and across whichever AI tool each of them uses. Not a link you send; a layer you all query.

02

Per-paragraph provenance

Every fact carries who said it, in which chat, and when. Answers come back attributed, so you can see and trust where they came from.

03

Contradiction detection

When two teammates’ AIs disagree, Follow surfaces the conflict instead of silently picking one — the contested points stay visible.

← PreviousHousing WorksNext →Greener Hours

Rishabh Salian — portfolio · 2026 · built in the open

  • Email
  • LinkedIn
  • GitHub
  • Behance