Section 0.3: AI Capabilities & Abram, the ABRAM Assistant

[!IMPORTANT] AI Assistant & Automation Notice AI features on ABRAM (including matchmaking, resume imports, document generation, and Abram's Action Plans) are tools to help streamline your workflows. AI outputs are generated probabilistically and are not a substitute for human judgement. Users are solely responsible for reviewing and approving all budgets, contracts, invites, and payments.

The ABRAM Network incorporates an integrated suite of artificial intelligence systems designed to streamline project intake, optimize crew composition, accelerate freelancer onboarding, and provide a conversational interface for managing production tasks.

These AI-driven tools leverage advanced language models, real-time availability mapping, and automated approval controls to keep resource allocation efficient while protecting your data and preventing system abuse.

1. Brief Intelligence (AI Brief Analyzer)

The AI Brief Analyzer (referred to in the interface as Brief Intelligence) acts as an automated intake coordinator, enabling project managers and producers to build structured project blueprints from creative briefs.

How it Works

• Brief Upload & Input: Users can drag and drop a brief document (PDF, DOCX, or TXT under 5MB) or type a manual project description (minimum 100 characters). Structured templates (Filmmaker, Marketing, and Creative) are available to guide input.
• Automatic Scoping: Once you trigger the analysis, the AI engine reviews your brief and automatically fills out:
  • Project Overview: Title suggestions, brief summaries, and primary creative industry.
  • Work Packages & Phases: Suggested phases of the project (e.g., Pre-Production, Production, Post-Production).
  • Required Roles & Skills: Key personnel slots needed, mapped to specific creative disciplines.
  • Deliverables & Tasks: Specific file, link, or milestone outputs, including target hours, priorities, and deadlines.
  • Schedule Constraints: Physical location requirements and specific shooting or production dates.
  • Technical Specifications: Software suites and hardware/camera equipment packages.
  • Estimated Project Scope: Suggested budget ranges based on project complexity and crew requirements.

Project Scoping Check

To ensure the draft accurately reflects your requirements:

• If the brief is clear, the AI drafts the project and presents the talent matching page immediately.
• If details are missing or ambiguous (such as unspecified locations or timelines), the system asks 3–5 quick clarifying questions to refine the details before saving.

For the full walkthrough, including budget estimate guidance, see the 2.1 AI Brief Analyzer guide.

2. Crew Matchmaking Suggestions

The matchmaking engine automates crew recruitment by evaluating candidate suitability for work package role slots. In the current phase of the platform, matchmaking suggests candidates from your internal roster only.

How Project Hours are Assigned

To suggest candidates, the platform calculates the hours needed for each role. It looks at three factors in order of priority:

1. Task-Specific Hours: The total hours assigned to tasks linked to a specific role.
2. AI Estimates: Project hour estimates automatically suggested by the AI based on the brief description.
3. Roster Split: If no specific hours are specified, the system splits the total phase hours evenly among the roles.

Match Suitability Criteria

Candidates are evaluated qualitatively based on how closely their profiles match your project scope. The platform evaluates:

• Skills & Experience: Proficiency in the required software, hardware, and creative disciplines.
• Project Track Record: Work history and successfully completed milestones.
• Availability & Location: Current calendar openings, travel preferences, and matching timezone.
• Collaboration Style: Preferred team dynamics and communication channels.
• Rate Compatibility: Alignment between the freelancer's day/hourly rates and your target budget.

Note: The interface lists specific Match Reasonings (strengths) and Concerns / Red Flags (such as rate limits or calendar conflicts) under each candidate.

For the full walkthrough, see the 4.2 AI Matchmaking Suggestions guide.

3. AI Resume Importer

The AI Resume Importer simplifies profile setup and skill tracking for freelancers, moving from manual entry to a single-action upload.

• Drag-and-Drop Interface: Freelancers upload their resumes (PDF, Word, or plain text files up to 10MB) directly from their profiles or onboarding screens.
• Autofill Capabilities: The AI extraction model parses the uploaded file and automatically populates:
  • Profile headline and professional bio.
  • Work experience history (company names, titles, durations, and descriptions).
  • Skill listings, categorized by discipline (e.g., Camera, Post-Production, Lighting).
  • Primary location and contact details.
• Review & Edit Stage: All parsed information is presented in a staging screen. Freelancers can adjust, remove, or approve the pre-populated values before saving them to their profiles.
• Onboarding Exemption: While standard profile updates consume AI credits from an organization's balance, resume parsing performed during the initial registration and onboarding wizard is completely free.

4. Abram, the ABRAM Assistant

Abram is the ABRAM Network's built-in AI assistant. You can open Abram two ways: as a slide-out side panel that stays with you while you work on other screens, or as a full-page chat for longer, focused sessions.

What Abram Can Do

• Answer questions and look things up: Ask Abram about your projects, schedule, budgets, invoices, and team capacity in plain language, and it will pull current answers with links to the relevant pages.
• Run multi-step Action Plans: Abram can carry out multi-step tasks — such as creating a project, sending invitations, or booking calendar time — but it always builds an Action Plan first and waits for you to click Approve before anything is actually created or changed. Nothing happens silently in the background.
• Generate documents: Abram can draft creative briefs, invoices, quotes, call sheets, and run-of-show documents into a preview canvas, which you can then export to PDF, Word, or CSV.
• Make direct updates: When you ask, Abram can update a project's status, budget, or schedule directly.
• Run script and screenplay breakdowns: Abram can kick off a script/screenplay breakdown as a background job, so you can keep working while it processes.
• Search the web: Abram can optionally search the web to help answer a question when the information isn't already in your workspace.

[!WARNING] Financial Guardrail Abram will never send a payment or approve an invoice on its own. Those actions always stay manual and must be completed by you in the financial area of the platform.

Suggested Prompts

Wherever you open Abram, it shows context-aware suggested prompts based on the page you're currently viewing, so you can jump straight into a relevant request. Suggested prompts don't cost anything to view or use.

Conversation Capabilities

• Talent Searches: Users can search the internal roster using normal language (e.g., "Find video editors in New York who are free next week"). Abram generates a list of candidates with direct links to view their profiles.
• Invite Dispatches: Ask Abram to invite selected freelancers directly to active projects or workspace rosters.
• Optimization Tasks: Ask Abram to evaluate team capacity, add skills, update profiles, or draft work packages.

Search Fallbacks

If a user specifies search criteria that return zero results (e.g., highly restrictive combinations of location, specific equipment, and expert rating):

• Abram dynamically relaxes search constraints rather than returning an empty page.
• It first removes physical location filters to search remote-capable team members, then widens accepted roles, and finally suggests top-rated crew with matching core skills.
• Abram clearly explains how it adjusted the search parameters in the chat response.

Safety Safeguards

To protect the integrity of the network, prevent spam, and maintain security, Abram enforces the following rules:

• Daily Invitation Limit: Users are limited to 10 external invitations per day. Once reached, Abram blocks further external requests and provides a countdown to when the limit resets.
• Duplicate Invitation Blocks: Abram cross-checks active invitations. If an invitation to the same email address is already pending for the project or platform, the action is blocked.
• Registered User Routing: If an entered email is already registered on the ABRAM Network, Abram blocks the external invite flow and redirects the producer to invite the user directly from the internal team roster.
• Off-Topic Protection: If users send Abram off-topic prompts (e.g., weather updates, general programming code, trivia), the system blocks credit consumption, explains its focus on project management, and suggests valid questions.

Action Plans and the Approval Gate

Before any transactional change — such as dispatching an invite email, booking calendar space, or creating a project — is executed, Abram generates an Action Plan in the conversation panel.

• Structured Summary: The Action Plan outlines the relevant details, such as the recipient's name, email, project, role, proposed rate, and target hours.
• Approval Gate: The action remains in a pending state until you click the green "Approve" button on the Action Plan card. Abram is programmed never to send emails, dispatch invitations, or execute updates silently in the background without this explicit click.

5. ABRAM Memory & Organization Brain

ABRAM's memory features give the platform's AI tools context so recommendations reflect how you actually work, instead of generic averages. There are two layers:

• ABRAM Memory: Your personal memory layer, managed in Settings → ABRAM Memory.
• Organization Brain: Your organization's shared memory layer, found inside Organization → Documents, which draws on project history, crew rosters, equipment, and past intake context across your whole workspace.

For a full breakdown of the features, search queries, and security permissions for both layers, see the standalone 0.4 Memory Guide.

6. Customizing Abram with Agent Skills

If you want Abram to consistently follow certain instructions, formatting preferences, or workflows, you can save your own Agent Skills. Go to Settings → Agent Skills to create named, reusable instruction sets that shape how Abram responds and acts for you, without having to repeat the same instructions in every conversation.

7. AI Credit Consumption & Optimization

AI features are metered and charged to your organization's AI credit balance.

Credit Usage Order

Credits are drawn from your organization's pools in this order:

1. Monthly Allowance (included in the subscription plan, resets monthly).
2. Trial Credits (issued during sign-up, expires after the trial period).
3. Purchased Balance (additional credits bought via Stripe, never expires).

How Credit Costs Work

Credit deduction is based on the actual complexity of the request and the amount of data processed, so costs vary from one request to the next — brief analyses, matchmaking suggestions, resume imports, script breakdowns, and Abram conversations do not have a single fixed price per action. Any cost figures you see in the product are approximate and reflect real-time usage, not a flat rate.

For a full explanation of how credits are calculated, the top-up packs available, and how to review your Credit Usage Log, see the 5.3 Billing, Ledger & AI Credits guide.

When Credits Run Out

If your organization's AI credit balance reaches zero, AI features are blocked and you'll see a prompt directing you to Go to Billing to top up or review your plan.

Smart Query Optimization

To minimize costs for repetitive tasks, ABRAM optimizes credit usage:

• If you ask follow-up questions or refine a project search within a short window, the system reuses recent results to keep costs low.
• This ensures that ongoing conversations with Abram or refinement queries to the Brief Analyzer remain cost-efficient.
• If a network interruption or timeout occurs during analysis, you will only be charged for requests that successfully complete.