AI that remembers
how your business works
Seshen-Prism keeps the reasons behind every decision — your records, your processes, and your AI all work from the same memory. Built for everyday business, not just engineers.
Proven Integration Wedges
Seshen-Prism, applied where memory matters most.
Seshen-Prism works with the systems you already use — it gives them a memory of how your business works, so answers come from your own history, not a guess. These are the areas where it's live or shipping first:
The first live build. Seshen-Prism layers your decisions and reasons onto Odoo — so the why behind every setup stays with your system, not the consultant who leaves. More platforms coming.
Records, approvals, and audit history with the reasoning attached — so nothing is just numbers without a story.
Your routine tasks run on autopilot, remembering what happened last time and why — fewer mistakes, less redoing.
A consultant configured your ERP years ago. Today the AI answers not just what you do, but why you do it — and shows its sources.
Try it: open a deployment and ask "Why is this set up this way?" — every answer carries a receipt you can check.
The Problem
Every ERP project
suffers from organised amnesia.
When a consultant wraps up an ERP project, they take everything with them — the reasons behind each configuration, the constraints that shaped every workflow, the warnings about what not to touch. What's left is a system that works, but that nobody fully understands why.
That isn't a people problem. It's an architectural one. A standard ERP rollout has no mechanism for keeping the institutional reasoning that turns stored data into knowledge you can actually act on.
Choices made for a specific business constraint become indistinguishable from defaults. Future changes quietly break things no one remembers protecting.
Every person who leaves takes undocumented context with them. Every new hire starts from zero — calling the same external partner at the same day rate.
Major ERP releases force a re-evaluation of every customisation. Without a memory of why those customisations exist, that becomes guesswork.
Off-the-shelf assistants give generic ERP answers. Without your specific decisions and history, that guidance is as likely to mislead as to help.
The Seshen Memory Engine
Five integrated components.
One institutional memory.
A purpose-built engine for keeping your business knowledge inside your ERP workflows — grounded in graduate research at Brunel University London.
Keeps the what (your records) separate from the why (your decisions) — and connects them, so a question finds the reasoning behind the record, not just keywords.
Every piece of knowledge is tagged before storage — fact, opinion, or official decision — so a passing suggestion never carries the same weight as a signed requirement.
Two agents work as a pair: one recalls your history and the reasoning behind it, the other handles the task at hand. Answers carry both institutional memory and current reality.
Outdated information is retired with a trace of what it replaced — while dissenting expert opinions are kept, because weak signals prevent costly mistakes.
Every answer shows its sources — which records and decisions it came from, and how confident it is. Verifiable, not just plausible.
Grounded in Research, Built for Practice
Seshen Labs' memory engine is grounded in graduate research at Brunel University London into organisational forgetfulness in AI workflows — validated through a PRISMA-compliant systematic review and Design Science Research methodology, then built for live ERP deployments.
Delivery is led by a small team spanning AI strategy, ERP implementation, and financial systems, with Prose AI as delivery partner. Every deployment feeds a growing body of knowledge that makes the engine sharper for the next one.
We're early-stage and building in the open — real deployments, not slideware. Book a call if that's the kind of partner you want.
Give your ERP a memory it keeps.
Seshen-Prism is live — metered gateway, isolated vaults, provisioning console. Tell us what you're building and we'll scope it.
FAQ
Straight answers.
Is the platform live or a roadmap?
Live. The multi-tenant metered gateway runs in production today — per-tenant keys, token metering, spend caps — and Seshen runs its own operations on the same stack. Tenant vault isolation and the provisioning console are live; subscription billing follows, with payment registration already running. Agencies can start on provisioned, isolated tenants now.
What does Seshen Labs actually do?
We give AI systems a memory of your business. Seshen-Prism keeps the reasoning behind every decision, so your knowledge stays with your business instead of leaving with a consultant. Our first live application is ERP on Odoo, with more platforms on the roadmap.
What is the Seshen memory engine?
It's an architecture, grounded in research at Brunel University London, that preserves the why behind every setup. It sorts knowledge by trustworthiness and pairs a history agent with a task agent, so every answer comes with its sources — not just a confident guess.
How much does it cost?
Every business is different, so we don't publish list prices. Book a call, we'll scope your project, and we'll give you a fixed quote — no surprises and no hourly meter running in the background.
How long does an implementation take?
For a fresh Odoo build, most implementations run about 6–10 weeks, versus a typical 12–16. Adding the memory layer to an existing ERP (BYO ERP) is scoped per engagement — book a call for a timeline.
Who is behind Seshen Labs?
Seshen Labs was founded by Vincent Bosire — MSc AI Strategy (Brunel), ACCA — with 20+ years across investment analysis, management accounting, and AI consulting in Kenya and the UK. The framework comes directly from his graduate research.