AI project intelligence for connected work

Your projects are connected.
Pentra sees what happens next.

Plan projects, map dependencies, predict delays, simulate changes and let Pentra help your team act before milestones move.

No credit card required.

ERP Implementation

Dependency cascade

Move one date.
See the entire cascade.

Every task knows what it waits for. Slip Vendor approval and watch the plan answer: which tasks move, which milestones follow, and whether float absorbs it before go-live does. Real scheduling, on a sample ERP plan.

On plan
days
Vendor approval
Oct 10
Downstream tasks
0 affected
Milestones
0 affected
Go-live
Nov 12
ERP ImplementationCurrent plan
Pentra

Vendor approval is on plan.

Go-live holds on Nov 12. Move the date to see what depends on it.

How Depentra works

Plan it. Understand it. Act on it.

Depentra models the connected project. Pentra understands it. One task, from the plan to the decision:

  1. Plan

    • Projects
    • Tasks
    • Milestones
    • Dependencies

    Vendor approval feeds Integration build and Security review, which feed testing.

  2. Understand

    • Impact
    • Risk
    • Critical path
    • Forecasts

    +10 days moves 7 tasks and 2 milestones. Go-live moves 5 days.

  3. Act

    • Simulate
    • Agents
    • Workflows
    • Approve

    Recover 5 days of the 10-day slip on Vendor approval: 5 days back, low risk, applied when you approve.

Pentra Intelligence

Pentra doesn't just answer questions.
It understands the plan.

Predict
Forecast every milestone from the plan itself, and see which ones are slipping before anyone reports it.Every plan
Explain
Ask Pentra why: the cause, the chain it travelled and who owns what, in plain words.Every plan, uses AI Credits
Simulate
What-If: try a delay or a change on a copy of the plan and compare it before you commit.Professional
Recover
Recovery options measured on the schedule, prepared by an agent, applied only when you approve.Professional
Project healthERP Implementation

Needs attention

Predicted go-live delay
+5 days
Primary cause
Vendor approval
Affected
7 tasks, 2 milestones
Potential recovery
5 days
Recommendation
Recover 5 days of the 10-day slip on Vendor approval

The forecast, cause and recovery are computed by the scheduling engine and use no AI Credits. Asking Pentra to explain them does.

Views

One project.
Every view your team needs.

Table, Gantt, Board and Dashboard are the same plan: change a date in one and every other view, dependency and forecast follows. Twenty views in all, from RAID to Budget.

ERP ImplementationGantt view
TaskOwnerStartEndStatusFloat
Requirements sign-offPriya ShahSep 14Sep 25Done-
Vendor approvalTom EllisSep 28Oct 10In progress-
Security reviewAna CostaOct 11Oct 22Not started9d
Integration buildKenji MoriOct 11Oct 26Not started5d
Data migrationLena BrandtSep 28Oct 31In progress0d
System integration testingKenji MoriNov 1Nov 4Not started0d
User acceptance testingSofia ReyesNov 5Nov 9Not started0d
TrainingUnassignedNov 5Nov 10Not started0d

Not started 4

Security reviewAna Costa
Integration buildKenji Mori
System integration testingKenji Mori · Critical path
User acceptance testingSofia Reyes · Critical path

In progress 2

Vendor approvalTom Ellis
Data migrationLena Brandt · Critical path

Done 1

Requirements sign-offPriya Shah
Project healthNeeds attentionGo-live forecast late
Tasks101 done, 2 in progress
Milestones2Go-live Nov 12
Critical path8tasks with no float
Plan100%
Procurement40%
Build35%
Test0%
Deploy0%

Dependency intelligence

Know what moves before you move it.

Four dependency types with lead and lag, on every plan. Depentra validates each link and refuses circular chains, then works out what they mean for your dates.

  • Critical path on the timelineStandard
  • Downstream impact and dependency chainsStandard
  • How a delay travelled, to its root causeStandard
  • Dependency conflicts and dependency riskStandard
  • Dependencies between projects, in ProgramsStandard
  • What-If: simulate a delay before it happensProfessional
FS Finish to startStarts when the other finishes.
SS Start to startStarts when the other starts.
FF Finish to finishFinishes when the other finishes.
SF Start to finishFinishes when the other starts.

Lead and lag on any link: start three days after, or two days before.

  • 5,000Linked tasks scheduled, float and critical path included, in under half a second. A test holds us to it.
  • 45Templates with ready-made views, fields, workflows, dependencies and dashboards.
  • 4Dependency types: finish-to-start, start-to-start, finish-to-finish and start-to-finish, each with lead or lag.
  • 1,100+Automated tests, from the scheduling engine you just tried to the database rules that keep projects private.

Needs attention

Know where to look first.

The schedule is checked all the time: late milestones, work that moved onto the critical path, tasks nobody owns. Findings arrive worst first, resolve themselves when the cause goes away, and cost no AI Credits.

Pentra adds what the numbers mean, and can prepare a recovery plan for your approval.

Needs attentionERP Implementation · worst first
  • Critical
    Go-live is forecast 5 days lateForecast Nov 17 against a plan of Nov 12.
    New
  • High
    Vendor approval now drives the go-live dateIt moved onto the critical path when Integration build ran out of float.
    New
  • Medium
    3 tasks have no ownerCutover rehearsal, Training and Hypercare.
    New
  • PentraUAT is the next milestone at risk: it starts after the slip lands. Want me to prepare a recovery plan?

AI Agents · Standard and Professional

Give Pentra a goal.
Let Agents handle the work.

An agent watches a project on a schedule or when tasks change. It reads the plan, explains what it finds, and on Professional it prepares a recovery plan, measured on the schedule.

  • Controlled. It proposes; a person approves before anything in the plan changes.
  • Contained. It sees only what the person it acts for can open.
  • Budgeted. A daily AI Credit cap per agent, and a quiet project costs nothing.
GoalFind what is threatening ERP go-live.
Risk agentERP Implementation · weekly and when tasks change
  1. ObserveReading tasks, dependencies and ownerswaiting
  2. AnalyzeMeasuring milestones against their floatwaiting
  3. AnalyzeExplaining the risk (AI)waiting
  4. PlanBuilding a recovery planProfessionalwaiting
  5. SimulateMeasuring the change on the scheduleProfessionalwaiting
  6. ProposePreparing it for your approvalProfessionalwaiting
Recovery plan ready: 5 days backWaiting for your approval. Nothing changes until you apply it.

AI Workflows

Automate the work around the work.

Rules that run when tasks change or on a schedule: set a status, assign, notify, raise a RAID entry, ask for approval. On Standard and Professional a step can ask AI to analyse the project. Date changes are proposed, never made silently.

  1. WhenOn a scheduleEvery Monday at 08:00
  2. ThenAsk AI to analyseProject healthEstimated 2–10 AI Credits
  3. ThenAsk AI to analyseWeekly status reportEstimated 2–10 AI Credits
  4. ThenNotify someoneThe project owners

Result Every Monday the owners get the week's status and what needs attention, before stand-up.

Programs · Standard and Professional

See the program before one project becomes the problem.

Roll projects up into a program, see each one's health from its own plan, and measure the dependencies that run between them.

Digital Transformation12 projects · sample program
  • 7On track
  • 3Needs attention
  • 2Off track
  • CRM Platform
  • Master Data (MDM)
  • E-commerce Relaunch
  • ERP Implementation
  • Data Warehouse
  • Identity & Access
  • Payments Upgrade
  • Mobile App
  • Contact Centre
  • Analytics Rollout
  • Network Refresh
  • Service Desk

Solutions

One platform.
Different ways of working.

Steer a portfolio of initiatives against one roadmap.

Program RoadmapFrom the Program Roadmap template
  • Customer portal refreshQ1 Initiatives
  • Payments platform upgradeQ1 Initiatives
  • Q1 reviewQ1 Initiatives
  • Data warehouse migrationQ2 Initiatives
  • Mobile app v2Q2 Initiatives
  • Q2 reviewQ2 Initiatives

Pentra Which initiatives threaten the Q3 steering committee dates?

See the Program Roadmap template

AI Credits

One AI Credit pool.
Every Depentra AI feature.

Pentra, Agents, AI steps in workflows and Create with AI all draw on the same credits. Each plan includes them per user, or per active paid member on a Team, every month.

  • Free100
  • Basic750
  • Standard2,500
  • Professional7,500
Pentra, Depentra's AI

Estimated AI Credits

  • Pentra: a quick answer~1–5
  • Pentra: a project analysis~2–15
  • Agent mode in Pentra~4–40
  • AI Agent run~2–8
  • AI step in a workflow~2–10
  • Create with AI~5–15

Estimates from what each feature may read and write; you are charged what a request actually cost, usually toward the low end.

No AI Credits

  • What-If: Delay Simulator and Scenarios
  • Recovery options for a threatened milestone
  • Critical path, dependency conflicts and chains
  • How a delay travelled (the root cause)
  • Impact Matrix and downstream cascade
  • Needs attention findings and early warnings
  • Project health scores

A Team shares one pool

Professional Team5 × 7,500 = 37,500AI Credits a month, shared

  • SarahPentra
  • JohnAgent
  • AlexWorkflow

Same organization pool

Workspaces and projects don't create new pools; they show where it went.

How AI Credits work

Pricing

Start free. Grow into Team.

Compare all plans

One person, one workspace. Billed annually; monthly billing is also available.

Free

Explore connected project planning

$0forever

100 AI Credits a month to try

  • 1 project, up to 2 people
  • Dependencies of all four types
  • Pentra to try it
Get started free

Basic

Plan and manage your work

$9per user / month

750 AI Credits a month

  • Timeline (Gantt) and Board
  • 5 projects, 200 tasks each
  • 1 workflow switched on
Choose Basic

Standard

Most popular

Understand connected work

$12per user / month

2,500 AI Credits a month

  • Critical path, impact and Programs
  • AI Agents and Create with AI
  • AI steps in workflows
Start 7-day free trial

Professional

Predict, simulate and recover

$19per user / month

7,500 AI Credits a month

  • Predict and simulate: What-If
  • Recover: agents prepare recovery plans
  • Agents across a workspace or organization
Choose Professional

Security and control

AI you can put near a real plan.

How Depentra protects your data
Your data stays in its organization
Every request is checked by the database itself, row by row: what you cannot open is never sent to you.
Access per project
Owner, editor, contributor or viewer on each project; guests see only the projects and views you share.
AI works inside your permissions
Pentra and agents read only what the person they act for can open, and every AI request is checked first.
AI proposes, people approve
Agents and workflows prepare changes; nothing in a plan moves until someone applies it, and every change can be traced.
A history of every change
Who changed what and when, on every task, with restore.
Payments by Stripe
Card details go to Stripe and never reach Depentra.

Questions, answered.

All pricing questions
Which plan should I choose?

Free to try Depentra on one project. Basic to plan and manage your own work on a timeline. Standard when you need to understand how the work connects: the critical path, downstream impact, Programs, AI Agents and Create with AI. Professional when you must predict and recover: What-If simulation, agents that prepare recovery plans, RAID, budget and resources.

Is Depentra priced per user or per workspace?

Per user, or per paid seat on a Team plan; never per workspace. Free is free; Basic, Standard and Professional are billed per user per month, monthly or annually. A Team pays the same price per seat, from 3 seats.

What are AI Credits?

What Pentra, AI Agents and AI steps in workflows spend when they ask a language model. Each plan includes AI credits per user, or per active paid member on a Team, every month: 100 on Free to try it, 750 on Basic, 2,500 on Standard and 7,500 on Professional.

Are AI Credits the same as tokens?

No. A request is charged what it actually cost to run, converted to credits, so you never count tokens. A quick question is about a credit; a whole project plan from a description is more.

How many AI Credits does each feature use?

Roughly: a quick question to Pentra ~1–5, a project analysis ~2–15, a summary or status report ~1–8, Create with AI ~5–15, an Agent mode request ~4–40, an agent run ~2–8 per project it interprets. Before a large request you see its estimate and are asked.

What doesn't use AI Credits?

Everything that is not a language model: planning, editing, every view, dependencies, the critical path, the Impact Matrix, What-If simulation, recovery options, health scores, Needs attention findings, imports, exports and workflows without an AI step. An agent run on a project where nothing changed costs nothing either.

See the problem
before it becomes the deadline.

Start with your project and let Depentra show you what happens downstream.