Competitive Analysis

[Zack-Rider]

Coyax — Competitive Research & Solution Strategy

Date: 2026-04-12
Author: Piyush
Purpose: Research companies building the same category of product as Coyax, understand what they do well, and define what Coyax should build and how.

---

What Coyax Is Building

Coyax is an AI-powered business operations platform for B2B companies (supply chain, procurement, logistics) with:

• Custom data tables per organisation (with relationships and views)
• Workflow automation engine (node-based, DAG execution, 10+ node types)
• AI agents (form builder, SQL agent, forecast, MDM)
• Third-party integrations (HubSpot, Slack, Gmail, Google Drive, Notion, SAP, Oracle)
• Document generation and parsing
• Chat interface for interacting with data

---

Part 1 — Competitive Landscape

1.1 BackOps AI

Category: AI-native operating system for supply chain
Funding: $6M Seed (June 2025) + $26M Series A (March 2026, Theory Ventures)
Founded by: Amazon and Apple alumni

What they do:
BackOps is the closest direct competitor to Coyax. They are building an "AI OS for global supply chains" — automating the back-office operations of logistics teams (carrier claims, shipment exceptions, customer inquiries, document collection).

Their two core products:

• AI Process Center — records how employees complete workflows today, then converts those patterns into automated actions
• Relay — an agentic automation engine that runs continuously across email, Slack, and portals, detecting and resolving issues automatically

Proven outcomes: 93% faster customer response, 60% time savings, 100% of eligible carrier claims filed automatically.

How their engine works:

• Process capture first — learn from human behaviour, then automate
• Agentic execution — the AI decides steps based on context, not a rigid flowchart
• Runs continuously, not on manual trigger
• Communicates via email and Slack — works in channels users already use

Why this matters for Coyax:
BackOps just raised $26M and is in the exact same market. Coyax cannot win by building the same thing. The differentiation has to be clear.

---

1.2 Magentic

Category: AI agents for procurement and supply chain
Funding: $5.5M pre-seed (July 2025, Sequoia Capital, First Momentum)
Founded by: Ex-OpenAI engineers and McKinsey consultants

What they do:
Magentic builds "Mages" — AI agents embedded in procurement teams that handle reconciling contracts, auditing supplier performance, identifying cost leakages, and reviewing supplier documents.

Business model: Pay-per-cure — customers only pay when actual savings are delivered. No software license fee. Early clients are recovering €8.5M–€17M per year.

How their engine works:

• Agents connect directly to systems of record (SAP, Oracle, procurement platforms)
• Core capability is document intelligence — reading unstructured contracts, invoices, BOLs and extracting structured data
• Human-in-the-loop for all approvals
• Deterministic actions (the AI finds the problem; the action to fix it is rules-based)

Why this matters for Coyax:
Magentic is specifically in procurement — a sub-vertical within supply chain. Coyax already has document-parser and document-creation nodes.

---

1.3 Relay.app

Category: AI-native workflow automation
Funding: ~$30M Series A

What they do:
Relay is a next-generation Zapier — built with AI as a first-class citizen. Operations and RevOps teams use it to build automations that combine structured logic (if/else, loops) with AI judgment (summarise, classify, extract).

How their engine works:

• Visual canvas with trigger → steps → actions (same pattern as Coyax's WorkflowEngine)
• Human-in-the-loop steps — pause, assign to a person, continue after they respond
• Every workflow run is persisted with a "Runs" dashboard
• AI steps are native nodes (not add-ons): summarise, classify, extract, draft reply

What Coyax can take:

• AI steps as first-class nodes (Coyax has Summarizer — needs Classify, Extract, Draft)
• Runs dashboard (Coyax's WorkflowEngine does not persist runs yet — this is a critical gap)
• Collaborative workflows where different team members own different steps

---

1.4 Lindy AI

Category: AI assistant / agent platform for business ops
Funding: ~$10M Seed

What they do:
Lindy lets teams create "Lindies" — AI employees that handle specific tasks: triage emails, schedule meetings, qualify leads, manage support, send follow-ups. The key differentiator is agent memory — Lindies remember context across runs.

How their engine works:

• Natural language workflow definition — describe what you want, Lindy builds the steps
• Agent memory: short-term (within a run), long-term (across runs), shared (across agents)
• Multi-agent coordination — one Lindy hands off to another
• 3,000+ integrations

What Coyax can take:

• Natural language workflow creation — "describe this in plain English, I'll build it" lowers the barrier for non-technical ops users
• Agent memory across runs — Coyax workflows are stateless today; persistent memory is very valuable for supply chain (track a supplier relationship over time)

---

1.5 Gumloop

Category: Visual AI workflow builder
Funding: Early stage seed

What they do:
Gumloop is a visual node-based workflow builder with AI as a native node type. Very similar product surface to Coyax's workflow feature. Users build automations combining AI steps with integrations.

How their engine works:

• Drag-and-drop canvas, same as Coyax
• Per-node testing — test a single node in isolation before running the full workflow
• Subflows — a workflow can call another workflow as a reusable block
• Cron scheduling built in

What Coyax can take:

• Per-node testing — click "test this node" to see its output. Dramatically speeds up workflow building.
• Subflows / reusable workflow blocks — template nodes from existing workflows
• AI Extract node — "give me a document and a schema, return structured JSON" (relevant for supplier invoices, BOLs)

---

1.6 Dust.tt

Category: AI agents for enterprise teams
Funding: ~$16M (backed by Sequoia)

What they do:
Dust builds custom AI assistants connected to internal company data — Notion, Slack, Google Drive, GitHub, Salesforce. Assistants can answer questions using real company knowledge and take actions via tools.

How their engine works:

• RAG (Retrieval-Augmented Generation) — AI answers questions by searching your indexed internal data
• Access control per assistant — legal assistant cannot see HR data
• Slack-native — assistants respond when @mentioned in channels
• Multi-agent architecture — one assistant delegates to a specialist agent

What Coyax can take:

• RAG over company data — connecting Coyax's AI chat to the org's own table data and documents would be a major upgrade to the SQL agent
• Access control per agent — important for multi-department enterprise use
• Company knowledge as product: "ask a question about your supply chain in plain English, get an AI-synthesised answer from your own data"

---

1.7 Relevance AI

Category: AI agent builder and workflow platform
Funding: ~$20M

What they do:
Relevance lets teams build custom AI agents and tools — mostly used by sales and ops teams for research, lead enrichment, content creation, and data processing. Has a "bulk run" feature for running an agent on hundreds of records.

How their engine works:

• Tool-first architecture — build reusable tools, compose them into agents
• Bulk run — run an agent on every row in a table (e.g. enrich all 500 contacts)
• Long-running agents that work for minutes to hours
• Agent marketplace — pre-built agents for common tasks

What Coyax can take:

• Bulk run on table rows — run a workflow on every row in a Coyax table. Very powerful for supply chain (check compliance docs for all suppliers, update all order statuses).
• Pre-built workflow template library — "Supply Chain Starter Pack" of 10 common workflows
• Tool-first abstraction — Coyax's node executors are already built this way; good validation

---

1.8 Airtable

Category: Database + workflow + AI platform
Funding: $1.36B raised. Well-established.

What they do:
Airtable is the most mature "flexible database for ops teams" product. Their data model is the closest to Coyax's table architecture — linked records, typed fields, multiple views of the same data, automations triggered by record changes.

Their data model (study this carefully):

• Base → Table → Record → Field (typed: text, number, date, select, attachment, linked record, formula, rollup, lookup)
• Views: Grid, Kanban, Gallery, Calendar, Gantt, Form — same data shown differently
• Linked records — click to link a record in Table A to a record in Table B (relational, but friendly)
• Automations triggered by: record created, record updated, form submitted, scheduled time, webhook received

What Coyax can take:

• Linked records UX — Coyax has relationships in the schema; making them as easy as Airtable's "link to field" is critical
• Views system — Kanban and Calendar views are high-value for ops teams
• Forms that write to tables — expose a shareable form URL; a supplier fills it in and a record appears in your table. Very powerful for onboarding external parties.
• Record-change trigger — "when a record is created/updated, run this workflow." This is the most-used automation trigger and Coyax does not have it yet.
• AI fields — a column whose value is auto-generated by AI (e.g. "Risk Score" column calculated from order history)

---

1.9 Retool

Category: Internal tools builder + workflow + AI
Funding: $145M (Sequoia, Series C)

What they do:
Retool is the most developer-centric platform — engineering teams build internal dashboards, admin panels, and CRUD apps by connecting directly to databases and APIs. Their workflow product is a node-based automation builder, architecturally similar to Coyax's WorkflowEngine.

How their workflow engine works:

• Node-based, same as Coyax
• Every workflow run is persisted with full step-by-step input/output logs
• JavaScript transform steps — write arbitrary code inside a workflow node
• Triggers: webhook, cron, manual, app event

What Coyax can take:

• Workflow run logs are non-negotiable — Retool's workflows store every step's input, output, duration, and error. Coyax must build this.
• Direct SQL query builder — Coyax's SQL agent already does this via AI; a manual SQL query builder for power users adds depth.
• Retool is developer-focused; Coyax is ops-focused — different buyer, less direct competition, but the feature patterns are the same.

---

Part 2 — Competitive Map

                    MORE AI / AGENT-NATIVE
                           ▲
              Lindy AI     │    BackOps AI ← DIRECT COMPETITOR
              Magentic     │    Dust.tt
                           │
 NON-TECHNICAL ◄───────────┼───────────────► DEVELOPER / ENTERPRISE
 (easy, ops-first)         │                 (powerful, eng-required)
                           │
              Gumloop      │    Retool
              Relay.app    │    Airtable
                           │
                           ▼
                    MORE WORKFLOW / AUTOMATION

Where Coyax sits: Upper-left-to-centre — AI-native AND accessible to non-technical ops teams. This is a strong position with less competition than upper-right (BackOps/enterprise) or lower-right (Retool/engineering).

---

Part 3 — Gap Analysis

What competitors have that Coyax does not yet:

Feature	Who has it	Priority for Coyax
Workflow run history (persisted logs)	Relay, Retool, n8n, Gumloop	HIGH — table stakes for any automation product
Record-change trigger ("when a record is created, run workflow")	Airtable, Relay	HIGH — most valuable trigger for ops teams
Cron / scheduled trigger	Gumloop, Retool, n8n	HIGH — needed for the "Api Trigger / Scheduler" on the whiteboard
OAuth "Connected Accounts" for integrations	All of them	HIGH — prerequisite for any new integration
Per-node test button	Gumloop, Retool	MEDIUM — speeds up workflow building significantly
Linked records UX	Airtable	MEDIUM — relationships exist in schema, need better UX
Forms that write to tables (external shareable URL)	Airtable	MEDIUM — powerful for supplier onboarding
AI Extract node (document → structured JSON)	Gumloop, Relevance	MEDIUM — Coyax has document-parser, needs to surface it as a workflow node
Bulk run (run workflow on every row in a table)	Relevance	MEDIUM — high value for supply chain bulk operations
Agent memory across runs	Lindy	LOW (now) — powerful but complex to build
Natural language workflow creation	Lindy	LOW (now) — future differentiator

What Coyax has that most competitors do not:

Feature	Why it matters
Domain-specific AI agents (form builder, MDM, SQL, forecast)	Not just generic LLM calls — purpose-built agents for operations
Human-approval node in workflows	BackOps and Magentic both emphasise this; not implemented
Document generation (not just parsing)	Generate BOLs, POs, invoices as workflow outputs
Org-specific dynamic DB schema (Sequelize)	Each customer has their own schema; deeply flexible

---

Build order — what to ship next

Phase 1 — Fix the foundations (no new features, fix what breaks trust)

1. Persist every workflow run to DB (WorkflowExecution table). Show run history in UI.
2. Add record-change trigger: "when a record is created / updated in this table, start this workflow."
3. Add cron trigger: "run this workflow every day at 9am."

Phase 2 — Unlock integrations

4. Build the "Connected Apps" screen with OAuth flow (use Nango or build a simple OrgIntegration table).
5. Ship one deep integration end-to-end: HubSpot contacts sync → stored in Coyax table → trigger workflow when new contact added.

Phase 3 — Deepen the product

6. Per-node test button in the workflow builder.
7. AI Extract node — paste a document, define a schema, get structured JSON output.
8. Bulk run — run a workflow on every row in a table.
9. Shareable form URL that writes to a table (for supplier onboarding).

Phase 4 — Differentiate with AI

10. RAG over org data — ask any question about your supply chain data in plain English.
11. AI column type — a field whose value is auto-generated by AI from other fields in the row.
12. Workflow template library — pre-built supply chain workflow templates.

---

Positioning recommendation for the founder presentation

Narrative to use:

"BackOps solves the logistics execution layer. Magentic solves the procurement document layer. Neither of them gives you the flexible data tables, the customisable workflow engine, and the AI query layer all in one. Coyax is the OS underneath — the place where all your operations data lives and where all your automations run. We are not competing with them; we are the platform that makes tools like them possible."

Why this works: It positions Coyax at a higher level of abstraction. BackOps and Magentic are point solutions. Coyax is the platform. This is how Salesforce positioned against industry-specific CRMs in the 2000s.

---

Sources

• BackOps AI website
• BackOps $26M Series A — BusinessWire
• BackOps $6M Seed — PR Newswire
• FreightWaves — BackOps coverage
• Magentic website
• EU-Startups — Magentic raises €4.6M
• Magentic launch blog
• First Momentum VC — Magentic write-up

---

Last updated: 2026-04-12 | Author: Piyush

[m13v]

the load-bearing part of this gap analysis is the ordering: run logs and the human-approval node ranked above more agent features. that's the pattern that separates ops automation teams adopt from the kind that demos well and stalls. everyone builds the agentic execution layer first, then hits the wall where no ops lead lets it touch SAP or HubSpot without a per-action approval gate and an auditable run history. once an agent writes to a system of record, the trust surface (what did it do, what's it about to do, can i stop it) is the product. phase 1 is right to land that first.