Nathan Weill here.
I'm the CEO @ Flow Digital, where we help companies unleash their full potential by strategically automating every inch of their workflows. Each week, I share hot AI and automation tips to help you move your business into the future successfully.
Turn One Piece of Content Into Video, and Back Again

Animated with Claude Motion
Three tools that convert your best content without a production team
Your best blog post or training doc took hours to write and gets read once. Turning it into video, or a recording into an article, usually means an editor, a freelancer, or a lost weekend. Video tools now do most of that work, and can even put the video step on autopilot.
Charts and reports that animate with your exact numbers: Claude Motion turns a prompt and your own report into a short animated video, with no generated footage or people. Change one figure and the rest stays put. It's in beta on Team and Enterprise plans.
Blog posts to video: Pictory turns scripts and text into videos with footage, voiceover, and captions. Its Zapier setup can turn new blog posts from an RSS feed into videos and send them to your social channels.
Recordings to articles: Descript turns a video or podcast script into a blog post in one click.
Choose Claude Motion when the numbers must be exact, Pictory for blog volume, and Descript when the recording comes first.
Editing hours no longer cap your content calendar. Start with one workflow: a new blog post triggers a draft video, and a person approves brand and accuracy before anything goes live.
Motion Graphics: Which Animation Tool Fits Which Job
Animated charts and explainers are covered above. When you need footage instead, three tools generate it, and each fits a different job.
Cinematic camera moves: Higgsfield generates footage with preset camera motion and character consistency, and it offers an API. Sources disagree on pricing, so check its pricing page before budgeting.
Editing control: Runway is the pick when a team needs to direct and revise a clip, and it has an API.
Lowest cost per clip: Kling is repeatedly called the best value, and its API bills by the second.
The tools also chain. Claude Motion exports to Higgsfield and Runway for polish, so you can build the exact-number version first and add visuals after. Have a person check every figure before anything goes live.
Let Airtable Build the Base for You
Omni 2.0 builds your base, and new controls show who changed what.

Drop in a messy file and Omni 2.0 sets up the base for you.
Teams lose days rebuilding the same tables and automations for every new process, and AI tools working without oversight add risk on top. Airtable's Fall Release tackles both, and three changes are available now.
Build the base from a messy file: Omni 2.0, now on every plan, imports any file, sets up tables and linked records, and writes automations from plain-language requests. It can also add AI to each record to summarize, classify, or enrich it.
Let your AI tools work in your data, with limits: the upgraded MCP lets Claude, ChatGPT, and Gemini read and write records. Admins approve each tool and restrict it to specific bases.
See who changed what: audit logs show whether a person or an agent took each action, and export as CSV or JSON.
Start with Omni on one process your team rebuilds often, then check the audit log to see what changed in Airtable. Not sure which process to pick? Talk with our Airtable experts and we will help you find where Omni saves the most setup time.
More in the release: apps, tighter agent controls, and what's next
Turn a base into an app or weekly deck: Canvas builds custom apps, dashboards, and presentations from a plain-language description, using your brand theme. It proposes a plan and waits for your approval before adding tables or fields. One base can feed a team app, an executive dashboard, and a self-updating weekly deck.
Know Canvas's limits before you plan around it: it is in open beta for AI Labs customers on all plans. It needs AI, AI Labs, and Anthropic or Bedrock enabled organization-wide, and it is not available to HIPAA-compliant or non-US data residency organizations.
Limit what agents can touch: agent permissions, coming soon, will control read and write access down to the field or record. An agent can never exceed the access of the person who connected it. A beta waitlist is open.
Bring your own agent: coming soon, this will connect your organization's existing agents to Airtable. No date has been given.
AI for healthcare teams: AI for HIPAA is in early access. Healthcare customers should contact their Airtable account manager.
Free monthly AI credits: every plan includes an allotment, so teams can try Omni without a separate budget line. See the billing details.
Want to see what ships next? Browse Airtable's 2026 roadmap.
Add AI as One Step in a Process You Already Run
We analyzed 6 months of builds and here’s what we learned

AI handles one step of five, while fixed rules, your systems, and a person handle the rest.
Many teams buy an AI tool, enjoy the demo, and still retype invoices by hand. The tool never connects to the steps around it.
Six months of client and sales calls show where AI earns its keep. 61% of the AI we built or are building runs as one step inside an automation workflow, most often in Zapier.
Here is how a document workflow runs:
Read the messy part: AI pulls key details from documents and emails, which is 49% of our live and in-progress AI tasks.
Check it with fixed rules: stable criteria stay in rules, which cost less and are easier to test.
Update the right system: approved data lands in your CRM or finance tool without retyping.
Send doubtful cases to a person: low-confidence results wait for a human before anything changes.
Pick one slow, repetitive step and build this around it. You get faster handling and a person in charge of exceptions.
See the full findings and what to implement in The AI Use Gap.
Wrapper or Built to Last? How to Tell Before You Buy
Know what you are paying for before the contract renews
Plenty of AI tools sold today are wrappers: a text box sitting on top of someone else's model, like Claude or ChatGPT. For simple jobs like drafting and summarizing, a teammate can get the same result directly in those tools, tuned to your business, for a fraction of the price. When the vendor changes its pricing or shuts down, the workflow your team built around it goes too.
A wrapper still makes sense for a quick pilot or a one-off task, where starting this week matters more than lasting. A built tool pays off once the work touches your customers, your records, or your money. It connects to your data, works inside your processes, and gets more useful the longer you run it.
Four questions in the demo show which one you are looking at.
Does it know your business? A strong tool already holds your customers, products, and rules before you type anything.
Can it do the work? The good ones update the record, send the invoice, or route the ticket.
Does it show its homework? Look for a log of what it did, what it used, and who approved it.
Would it still be useful without the AI? Real value comes from the connections and rules around the model.
If a vendor cannot answer all four with your own data on screen, you are likely buying a wrapper.
Where a wrapper earns its place
Wrappers are the fastest way to test an idea. You can start in days, the monthly cost is low, and walking away is easy.
Fast to start: a pilot can run this week with no integration work.
Low commitment: monthly pricing means a bad fit costs little.
Good for one-off tasks: drafting, summarizing, and brainstorming need no connection to your systems.
The tradeoffs show up with scale. Competitors can copy the same product in a weekend, the tool starts from scratch each session, and quality depends on a model vendor you do not control. It also cannot change anything in your CRM, accounting system, or inbox, so a person still does the last step by hand.
Where a built tool earns its place
A built tool carries your business rules with it, so results stay consistent across a team of 5 or 500.
Work gets finished: records update, approvals route, and follow-ups go out without copy and paste.
Safer with sensitive data: access limits, logs, and approvals are part of the design.
Harder to replace: your process lives in it, and the value compounds over time.
The cost is upfront effort. Expect weeks of setup instead of days, a higher price, and a need for clean data and a named owner on your side. Skip it for a task you do twice a month.
A ten-minute test for your next demo
Ask for three things before any contract:
Use your real data: have the vendor run a live example with your own customer or order records.
Ask what happens when the AI model underneath changes: a built tool has a plan, and a wrapper often does not.
Compare the price: check what the same task costs in a direct subscription to a major AI model.
Where a person stays in the loop
Keep a human approving anything that moves money, goes to a customer, or changes a contract. Review a weekly sample of finished work for the first month, then monthly once results hold steady.
See how this plays out in practice in our client case studies.
Smart Words of the Week
— Warren Buffett
Each story this week asks the same question of a different tool: what do you get back for what you put in? Ask it of the wrapper in the demo, the AI agent working in your Airtable base, and the video your team approves before it publishes.
![]() |
Stay in touch with our team from anywhere.



