Security work, brought together.
ERI is the workspace where security teams and AI agents carry out assessments, reviews and investigations together.
What ERI gives your team
Isolated working environments
Each Cell is a container workspace with private storage. Agents and shells run inside it, beside the files they use.
Proof stays attached
Findings keep their requests, responses, scripts and screenshots, each with its recorded bytes and fingerprint.
Review another person can trust
Independent review is recorded for the exact version. Authors cannot confirm their own finding or retest.
Why ERI
Agents can do more of the work. Someone still has to stand behind it.
Scope in a document, work in a terminal, evidence in a folder. ERI keeps them in one record.
Work spread across tools
Work kept in ERI
F-07 confirmed · 3 evidence files · scope rev 3
How it works
From scope to something you can stand behind.
Four stages, one connected record. The context travels with the work.
Define the work
Set objectives, targets, exclusions, timing and supporting material. Revise the scope as the engagement changes.
Work in a Cell
Open shells, files and your chosen agent inside an isolated working environment.
Keep the proof
Connect findings to requests, responses, scripts, output and screenshots.
Review and report
Review the recorded work and produce a report suited to the engagement.
Who it is for
For the people who answer for the work.
Security teams, consultancies and the reviewers who sign off their work. Each works with the AI agents they prefer, and people stay in charge of what is delivered.
Can we work with our own agents?
Yes. ERI is built around people working with the agents they choose.
How do we start?
Request a demo. We will walk through ERI with the kind of work you do, and talk about whether a pilot makes sense.
Product
Built around the engagement.
Not around a chat window. Explore each stage of the record.
Scope
Keep the working context current
Record the objective, targets, exclusions, testing window and supporting material. Each revision is kept and supplied to connected workspaces.
- Revisions kept, never overwritten
- Exclusions stated next to targets
- Material attached to the engagement
Cell
Work inside an isolated environment
Open agent and shell sessions beside the files they use, in a container workspace with its own private storage.
- Agent and shell sessions side by side
- Uploads and safe file previews
- Reattach to running sessions
Evidence
Read the retained evidence
Accepted evidence keeps its recorded bytes and fingerprint, along with who recorded it, from which Cell and under which scope revision.
- Bytes and fingerprint retained
- Safe, read-only viewer
- Cited by the findings it supports
Review
Follow a finding back to its proof
Read the issue, the reproduction and the impact, then follow the review and retest history for each version.
- Independent, version-specific review
- Authors cannot confirm their own work
- Retest recorded after remediation
Report
Produce the report from recorded work
Bring scope, summaries, technical write-ups and evidence into one report, with the sections and branding your client expects.
- Configurable sections
- Client-facing branding
- HTML, PDF, Word, Excel or CSV
- Network controls you can read
- Configured, applied and observed settings for each Cell.
- Scout, the assistant inside ERI
- Explains the work and prepares prompts within each user’s permissions.
- Five export formats
- HTML, PDF, Word, Excel and CSV, with client branding.
- Deployment that fits
- Hosted today. Your environment by qualification.
Teams and agents
One engagement. Different responsibilities.
People and agents each work in their own lane, and every step is recorded.
Agent
Lead agent
Plans the work inside an authorised Cell, keeps the board moving and hands tasks over.
- Can
- Work in its Cell, save evidence, draft findings
- Cannot
- Confirm its own findings
Coordination between lanes is optional. Agent provider sign-in and usage rules still apply.
Lead agent record
Example application review · Demonstration data
- Read scope revision 3Scope
- Planned three tasks on the main board
- Handed endpoint mapping to the support laneHandover
- Saved request-01.http as evidenceEvidence
- Drafted finding F-07 for reviewFinding
Agent
Supporting agent
Takes delegated tasks in its own lane and reports back to the lead.
- Can
- Work on delegated tasks, attach output
- Cannot
- Change the scope or plan
Coordination between lanes is optional. Agent provider sign-in and usage rules still apply.
Supporting agent record
Example application review · Demonstration data
- Received task: map in-scope endpointsHandover
- Ran check.py inside the Cell
- Attached capture-01.pngEvidence
- Reported back to the lead lane
Read-only
Watcher
Follows progress and raises questions without changing the work.
- Can
- Read the board, raise questions
- Cannot
- Run commands or edit records
Coordination between lanes is optional. Agent provider sign-in and usage rules still apply.
Watcher record
Example application review · Demonstration data
- Followed the main board
- Asked whether the payment provider is in scopeQuestion
- Question answered by scope revision 2Scope
Person
Reviewer
A different person makes version-specific decisions on findings and retests.
- Can
- Confirm, reject or request changes
- Cannot
- Review work they authored
Coordination between lanes is optional. Agent provider sign-in and usage rules still apply.
Reviewer record
Example application review · Demonstration data
- Opened F-07 version 1Finding
- Requested clearer reproduction stepsChanges
- Reviewed version 2 and its evidence
- Confirmed F-07 (version 2)Confirmed
- Retest scheduled after remediation
Bring one real engagement.
We will walk through how your team could scope the work, use its agents and produce a result another person can check.