In stealth · Founded 2026

Agents are replaceable. Memory is not.

We're the shared memory and identity layer that sits under a company's AI agents — so every agent knows who's who, remembers what happened, and can prove where each answer came from, all inside the company's own systems.

30-second thesisSkim → then go deep

The problem

Agents run on scattered, stateless, unsourced data.

The engine

One layer resolves identity, remembers with provenance, and governs — underneath any agent.

The edge

The only platform native across all six runtime capabilities.

Traction

Signed design partners in regulated verticals; building toward production.

The ask

Raising a pre-seed to convert the signed pipeline.

MiddyMind is the proprietary entity-resolution and memory engine AI agents are built on and run on. Every entity across your systems, resolved. Remembered with provenance. Served to any agent, per turn, inside your own tenant.

Customers bring their own agents - authored in plain language or as their own code. MiddyMind supplies resolution, bitemporal memory, runtime governance, and a live improvement loop the agent consumes but never has to build.

Status
Stealth · pre-seed
Codebase
Monorepo (uv workspace) · in build
Verticals in build
EmpowerReg · ClearedAI
Target runtime
Fly.io · in-tenant by default
Applying
MSFT for Startups

TL;DR

The entity-resolution and memory engine under any agent — governed, bitemporal, in-tenant. In stealth, building across 2 regulated verticals. Raising pre-seed.

  • 2
    Verticals in build
    EmpowerReg · ClearedAI
  • 6 / 6
    Runtime capabilities designed native
    Only platform scoping Native across all six
  • ≈ 90d
    Target design-partner cycle
    Signed → in-tenant → governed agent
Entity-generic resolutionBitemporal provenancePer-turn Signal BriefIn-tenant federationReversible mergesRTBF cascadeAutonomy tiersLive-eval loopFederated reward modelBring-your-own-agentSigned usage meteringEntity-generic resolutionBitemporal provenancePer-turn Signal BriefIn-tenant federationReversible mergesRTBF cascadeAutonomy tiersLive-eval loopFederated reward modelBring-your-own-agentSigned usage meteringEntity-generic resolutionBitemporal provenancePer-turn Signal BriefIn-tenant federationReversible mergesRTBF cascadeAutonomy tiersLive-eval loopFederated reward modelBring-your-own-agentSigned usage metering
§ 01

The problem

Companies are deploying agents on top of scattered, stateless, unsourced data - and getting confident, ungoverned, unprovable answers at scale.

The failure is structural and lives below the model. It recurs for every agent a company ships.

Failure 01

Fragmented identity

Signal about any entity is scattered across many systems; an agent sees slices.

Failure 02

Amnesia

Agents are amnesic across sessions and channels; what one learned or promised, the next never sees.

Failure 03

No provenance

Answers carry no source; when one is wrong, no one can say which system it came from or when it was last true.

Failure 04

Governance re-litigated

Residency, consent, retention, audit - renegotiated per source and per project instead of enforced once, underneath.

§ 02

The engine

Five primitives. Each a live engine module. The agent is pluggable and lives above this line; everything that makes it useful lives below.

  1. 01

    Entity Resolution

    identity/

    Resolves any entity type across sources.

  2. 02

    Memory Fabric

    memory/

    Bitemporal (valid-time + transaction-time) store of typed, sourced facts. Now legible and self-improving: correctable Memory Ledger, nightly Consolidation, orthogonal scoping, standardized recall benchmark.

  3. 03

    Signal Brief

    conductor/signal_brief.py

    A compact ~200-token typed snapshot.

  4. 04

    Governance

    governance/

    In-tenant federation, per-fact attribution, per-purpose consent, three autonomy tiers. Hard cost caps with model-independent, cost-tiered routing and a budget breaker.

  5. 05

    Live Eval & Improvement

    eval/, learning/

    “The benchmark is yesterday.” Nightly sampling of real interactions.

§ 03

The Conductor

The spine that ties the five primitives together. Every inbound message flows through one ordered lifecycle. Swap the model and the resolved entity graph compounds underneath.

conductor/ · per-message lifecyclelive
identitysignal_briefroutecost capagent callautonomywrite-back
theneval · audit · replay

Two hooks for the agent - read a Signal Brief on demand, write its outcome after. No plumbing. Omnigent swaps Claude / OpenAI / Gemini behind an identical brief.

§ 04

Verticals in build

Two verticals in active build — both in healthcare — with a third planned to prove the substrate is entity-generic. Each brought its own agents and domain; MiddyMind supplied everything underneath.

§ 05

Bring your own agent

Two authoring modes, one engine binding. Both select an entity type and connectors, set policy (autonomy tier + cost cap + consent purpose), and deploy onto the harness in-tenant.

Primary on-ramp

Describe it

A natural-language description is turned into a readable AgentClient subclass by a.

Advanced

Bring your code

Claude Agent SDK, LangGraph, a container image, or a custom HTTP endpoint.

§ 06

In build

In stealth. Verified in code, nothing yet in production. Raising a pre-seed round and applying to Microsoft for Startups to underwrite the next twelve months of in-tenant infrastructure.

For VC reviewers: a defensible substrate — entity-genericity, correctable Memory Ledger, closed eval loop — under whichever agent runtime wins.

For cloud-fund reviewers: in-tenant deployment across three sovereign channels, with cost-tiered routing and a per-tenant budget breaker governing consumption.

2[1]
Signed design partners (regulated verticals)
16[2]
Seeded agents across 2 client packages
53+[3]
Registered tools (ClearedAI + VC)
6/6[4]
Runtime capabilities native — Section 07
Agents seeded[2]16 / 2 packages
ClearedAI · 10
VC Copilot · 4
EmpowerReg · 2
Runtime capabilities[3]6 / 6 native
Bitemporal memory
Correctable ledger
Nightly consolidation
Orthogonal scoping
Cost-tiered routing
Budget breaker
Tools registered[4]53+ total
ClearedAI grounding34
VC diligence19
Recall benchmark[5]Internal · synthetic 500-item corpus · pre-production
0.94
Precision
0.91
Recall @ k=5
180ms
Latency p95

Fixed 500-item synthetic corpus · run nightly against Memory Ledger head · reproducible eval.

Assumptions & definitions

  1. [1] Design partners = signed MNDA + written scope; as-of current build cycle.
  2. [2] Agents = orchestrator + specialist roles wired in code with tool bindings; not headcount.
  3. [3] Native = implemented in-repo against § 07 capability list; no external SaaS dependency.
  4. [4] Tools = registered callables in the tool router with schema + policy; deduped across packages.
  5. [5] Benchmark figures are internal-eval, pre-production; methodology in § 12.

Stealth-appropriate signals · no revenue disclosed · no logo use without partner sign-off · full definitions & as-of dates in § 12.

§ 07

Competitive analysis

Six runtime capabilities, six categories

Every agent platform must own the six columns end-to-end. We scored the closest neighbors in each category against the same runtime bar we hold ourselves to.

“Coverage” counts only Native marks — Partial means the piece exists but ships as a library, service boundary, or manual step rather than a governed primitive. Hover any cell for the definition.

Closest neighbor

Zep / Graphiti covers 4 of 6 capability dimensions. Strongest single competitor, still not a runtime: no entity-generic resolution, no closed eval loop, no bring-your-own-agent harness. No rival — including the doc-AI acqui-hires, the RPA incumbents, and the cloud-ops / CNAPP tools (Wiz, Orca) — is native across all six.

MiddyMind

This engine

Entity-generic
Bitemporal
Per-turn brief
In-tenant
Live-eval
BYO agent
§ 08

Market opportunity

Sized top-down from analyst forecasts, bottom-up from the buying centers we already sell into. Numbers are dated, sourced, and footnoted — no round-number storytelling.

The thesis: agent spend is real and about to overshoot; a governed memory + orchestration substrate is what turns the 40% cancellation rate into deployed revenue.

$2.52T1
TAM — worldwide AI spend, 2026
Gartner forecast, +44% YoY. Infrastructure alone adds $401B as providers build out AI foundations.
$1.3T2
SAM — agentic AI IT spend by 2029
IDC: agentic AI to exceed 26% of worldwide IT spending by 2029 — the budget line MiddyMind's substrate sits under.
$234B3
SOM wedge — enterprise app spend at agentic-arbitrage risk
Gartner: seat-license SaaS revenue exposed to agentic replacement between now and 2030 — the governance + memory layer decides who captures it.
$12.9B4
Adjacent pull — CNAPP by 2030
Dell'Oro: 22% CAGR; Wiz and Microsoft neck-and-neck for share lead. Cloud-ops buyers already fund the in-tenant control plane we extend to agents.
Why now — catalysts
  • Deployment curve40% of enterprise apps ship task-specific agents in 2026, up from <5% in 2025.5
  • Failure mode = our wedge>40% of agentic AI projects will be cancelled by end of 2027 — cost, governance, unclear value.6
  • Regulated buyersGlobal GRC software: $65.2B in 2026, 12.2% CAGR — the budget our EmpowerReg + ClearedAI packages price into.7
Methodology — how each figure was derived
TAM
Top-down. Gartner worldwide IT-spend forecast (Jan 2026), global, calendar 2026. Definition: end-user spending on AI hardware, software, services, and infrastructure — reported as-is, no re-cut.1
SAM
Top-down. IDC agentic-AI share of worldwide IT spend, global, forecast year 2029. Definition: budget allocated to autonomous / goal-directed AI systems (runtimes, orchestration, memory, governance) — the layer MiddyMind's substrate sits under.2
SOM
Bottom-up wedge. Gartner "agentic arbitrage" estimate of seat-license enterprise-app revenue displaceable by agents, global, cumulative 2026–2030. Read as the pool of budget in motion, not the pool we capture — capture depends on which stack the deployed agents run on.3
Adj.
Dell'Oro CNAPP forecast, global, 2030. Definition: cloud-native application protection platform revenue (CSPM, CWPP, CIEM, KSPM). Included as the adjacent buying center — cloud-ops and CISO budgets — we extend into with in-tenant agent governance; Wiz + Microsoft cited as Q3 2025 share co-leaders.4
Catalyst
Analyst-reported adoption + failure-rate percentages (Gartner, Aug 2025 / Jun 2025), global, enterprise segment. GRC figure from Technavio, global, 2026 with CAGR to 2029. Used qualitatively to date the timing thesis, not to size revenue.5,6,7

Rules we followed: primary analyst releases only; every figure dated and geo-scoped; forecast years quoted as published (no interpolation); no vendor-sponsored TAMs; MiddyMind revenue projections are not derived by applying a capture rate to these pools — they're built bottom-up from signed design-partner ACVs and disclosed separately.

  1. Gartner, Worldwide AI Spending Will Total $2.5 Trillion in 2026 (Jan 15, 2026). source
  2. IDC, Agentic AI to Dominate IT Budget Expansion… $1.3T in 2029 (Aug 26, 2025). source
  3. Gartner, $234 Billion in Enterprise Application Software Spend Is at Risk from Agentic AI (Jul 1, 2026). source
  4. Dell'Oro Group, CNAPP to Reach $12.9B by 2030… Wiz and Microsoft Neck and Neck (Jan 13, 2026). source
  5. Gartner, 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026 (Aug 26, 2025). source
  6. Gartner, Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 25, 2025). source
  7. Technavio via BusinessofGRC, Global GRC market $65.2B in 2026, 12.2% CAGR through 2029. source
§ 09

Defensibility

Four things competitors cannot copy

MiddyMind is the entity-resolution and memory layer under the agent — complementary to the runtime, identity, and memory tools you already run. Read for two reviewers: VC partners underwriting a defensible substrate, and cloud fund leads underwriting governed in-tenant consumption.

  1. 01

    Entity-genericity

    One resolution engine for customers, devices, applications, ventures — a surface neither agent runtimes nor CNAPP/cloud-ops incumbents (Wiz, Orca) touch.

  2. 02

    Governance in-tenant

    The runtime enforces residency, consent and audit inside the customer's perimeter — the substrate cloud programs can underwrite as sovereign consumption.

  3. 03

    Correctable Memory Ledger

    Bitemporal ledger with nightly consolidation, orthogonal scoping, and a standardized recall benchmark — graded corrections feed a federated reward model.

  4. 04

    Cost-tiered routing + budget breaker

    Model-independent routing with hard per-tenant cost caps — the governance surface cloud funds care about, and a moat no memory or RPA vendor ships.

§ 10

How to buy

Deployment postures + pricing shape

MiddyMind is a PaaS you run inside your tenant. Pick a deployment posture below; the pricing shape follows.

Tenant isolation via Postgres row-level security on a per-request tenant GUC. The runtime federation guard refuses to start if the DSN looks like a vendor-owned shared cluster.

Tier 01

Shared-managed

Startups, non-regulated pilots

Our cloud.

Tier 02

Dedicated-in-VPC

Default · regulated enterprise

Runtime federation guard refuses to start against a vendor-owned shared cluster.

Tier 03

Air-gapped / sovereign

Government · defense

Offline install, signed traces only for remote diagnostics.

Economic buyer

Who buys, and what they pay for

Customers bring their own agents; MiddyMind is sold embedded under partners like EmpowerReg and Cleared AI, who carry it into the regulated buyer. The partner signs the commercial contract; MiddyMind is the substrate underneath.

EmpowerReg

The post-market safety lead

At a Class II/III medical device maker — owns MDR/QMSR obligations and signs for the safety-intelligence stack that EmpowerReg delivers.

Cleared AI

The compliance owner

Shipping a regulated software application — owns the artifacts a regulator or enterprise buyer will sign, and signs for the assurance package that produces them.

MiddyMind

Memory + governance substrate

EmpowerReg

Regulated SaaS partner

End customer

Device maker / hospital / regulator

MiddyMind runs underneath; the partner owns the customer relationship.

Pricing shape

PaaS with a design-partner phase

You rent the engine, we run it inside your tenant. Two co-designed usage counters are metered in the runtime today.

Early design partners get the platform floor waived in exchange for a revenue-share agreement on downstream revenue their MiddyMind-powered agents generate. Both sides win as the substrate compounds.

What you pay forDesign partnerNowStandard PaaSAt GA
Platform floorWaived. No annual fee during the design-partner phase.Annual fee by deployment posture — shared, in-VPC, or air-gapped.
Usage — resolved entitiesNot metered. You keep everything you build.Metered per entity under management.
Usage — agent-turnsNot metered.Metered per turn served.
Revenue shareYes. Agreed % of downstream revenue, fixed term.None. Floor + usage only.
CommitmentNamed design-partner agreement, joint roadmap, quarterly reviews.Annual contract, standard SLAs, self-serve expansion.
Billing evidenceSigned, tamper-evident cost ledger emitted by the runtime — verifiable on your side without breaking tenant federation.

Why this shape: design partners take real risk with us early, so we take revenue-share risk with them - no platform bill until their agents earn. At GA we drop the revenue share and switch to a transparent floor + two usage counters.

Illustrative: at a 20% share, a partner earning $1M through MiddyMind-powered agents returns $200k — growing as the substrate compounds.

Rejected: per-seat (punishes the “substrate fades” outcome), per-connector (taxes the wedge), pure per-token (positions us as a model reseller).

§ 11

Go-to-market plan

Three sequenced phases: convert the signed sovereign pipeline, template it, then open the standard PaaS with a federated loop.

  1. Phase 01Now - 6mo

    Convert the signed sovereign pipeline

    Ship in-country production deployments through EmpowerReg's regional partners.

  2. Phase 026-18mo

    Templated sovereign verticals + Ship Gate benchmarks

    Package each jurisdiction as an eval-backed template so new regulators deploy in days.

  3. Phase 0318-36mo

    Standard PaaS at GA + federated sovereign loop

    Convert design partners to floor-plus-usage pricing.

§ 12

Investors & funders

Sovereign-grade memory + governance, with a signed nation-state pipeline via EmpowerReg. Three regional partners — Network Partners (US), Excopan (Turkey), Mandala (LATAM) — are already fronting device regulators, hospital systems, and ministries of health in a $450B+ safety-intelligence market.

Written for two reviewers: VC partners underwriting a defensible substrate, and cloud program leads underwriting sovereign / in-country hyperscaler consumption. Traction, sovereign channels, market, timing, and sequenced GTM below.

Traction — defined, dated, sourced

Each metric below states its as-of date, an explicit definition of what counts, and a target with a confidence band. Nothing here is a projection dressed as a fact.

2
Verified
Signed design partners
As of Q3 2026
Definition
Countersigned design-partner agreement with a named executive sponsor and an active technical integration channel. Excludes LOIs, warm intros, and pilots without a signed contract.
Target / confidence
Target 6 by Q2 2027 · confidence high (3 in advanced conversation).
Source
EmpowerReg, Cleared AI.
3
Verified
Sovereign / nation-state channels engaged
As of Q3 2026 · via EmpowerReg
Definition
A regional partner with an existing book of business into a national regulator, ministry of health, or public health-safety authority — reached through EmpowerReg's countersigned reseller agreements.
Target / confidence
First in-country deployment target Q1 2027 (US) · confidence high; Q2 2027 (Turkey, LATAM) · confidence medium.
Source
Network Partners (US), Excopan (Turkey), Mandala Group (LATAM).
6 / 6
Verified · see Section 08
Runtime capabilities designed native
As of current build
Definition
Native = built as a first-class, governed primitive in the product (not SDK glue or a sibling service). Scored across the six-capability matrix in Section 08.
Target / confidence
Third-party benchmark publication vs Zep / Mem0 / Cognee targeted Q4 2026 · confidence medium (dependent on benchmark harness).
Source
Only platform scoping Native across all six dimensions today.
≈ 90 days
Target range 60-120 days
Target design-partner cycle
Planned across current 3 partners
Definition
Signed agreement → in-country VPC deployment → first governed agent running against real customer data with per-fact provenance and eval instrumentation.
Target / confidence
Target 45 days once vertical templates ship (H2 2027) · confidence medium.
Source
Reference architectures are reused across partners in the same jurisdiction.

Confidence · High = signed / observed · Medium = active pipeline · Low = plan of record only.

Sovereign pipeline via EmpowerReg

EmpowerReg is a US-based global health safety intelligence company already selling into device makers and hospitals through an AI-native SaaS platform for safety, risk intelligence, regulatory / compliance workflows, and post-market surveillance. MiddyMind is embedded as the memory + governance substrate under that stack — which means EmpowerReg's three regional partners double as MiddyMind's sovereign-tenant channel from day one.

United States
Network Partners Group

FDA-facing device makers + hospital systems.

Türkiye
Excopan

Turkish Medicines and Medical Devices Agency (TİTCK) ecosystem.

LATAM
Mandala Group

ANVISA (Brazil), COFEPRIS (Mexico), INVIMA (Colombia) and adjacent authorities.

Why now
01

Sovereign AI procurement is compressing

Ministries and regulators are standing up in-country AI mandates now.

02

Agents commoditize on a 12-18mo clock

Durable value moves to memory, governance, and eval underneath.

03

The closed eval loop is buildable

Federated learning on graded corrections is shippable today.

The ask
VC / Angel

Pre-seed round — 18-month runway

Funds the engineering to convert the signed sovereign / design-partner pipeline into.

Cloud program

Azure — startup

Underwrites in-country tenant infrastructure across every EmpowerReg regional.

The compounding argument: agents are replaceable, memory isn’t. Every source a customer connects, every correction they grade, every entity they resolve makes the substrate more valuable - to them and to every partner on it. That’s the asset both VC and cloud capital are underwriting.

End of brief · The ask

Agents are replaceable.
Memory is not.

If you’re an investor, a design partner shipping agents into a regulated tenant, or a program office evaluating in-tenant AI infrastructure — we’d like to hear from you. Everything else can wait until we’re out of stealth.

Under NDA · signed traces only
anu@middymind.ai

Every intro reply includes a signed traces bundle keyed to your deployment posture. Standard reply window: 24 hours.

For
Investors
Pre-seed round open
For
Design partners
Rev-share, floor waived
For
Program offices
Sovereign / regulated
For
Cleared talent
Selective hiring
MiddyMind AI · In stealth · Founded 2026Back to top ↑