Ops Digest #9: Using AI safely at the process level
September 13, 2023
May 25, 2023
This is a preview of Ops Digest, our monthly newsletter for operations teams. This month, we're talking about practical approaches to bringing to power of generative AI like GPT into your everyday processes. Subscribe and receive this newsletter in your inbox every month!
AI safety is top of mind in the tech industry right now. Look no further than recent Congressional hearings, where OpenAI CEO Sam Altam urged lawmakers to get going on regulating AI. As organizations all over are considering how to bring AI to bear on their next phase of digital transformation, questions linger as to how to do it, and if to do it—all because using AI safely is paramount.
It’s all too easy to get sucked into the AI hype versus AI doomsday extremes. It’s just more nuanced than that, and we have to understand that nuance if we're going to take advantage of the benefits AI has to offer while mitigating the risks.
The key is understanding that AI is fundamentally just a tool (or a set of tools), just like computers, the internet, cloud computing, and automation. As with any tool, you have to learn what it’s for, how it works, and how to use it safely and effectively.
Tools are things we use to get things done. And the people in large organizations who “get things done” tend to be operations professionals. Those “things”—that is, business processes—are ripe for improvement via automation and AI, especially things like process adoption and legal matter intake.
Ensure human-in-the-loop design, which creates transparency and enables people to see how an AI tool is working and step in if need be.
Use it for narrowly defined applications (even with powerful generative AI like GPT).
Understand the concerns germane to your vertical (e.g., there are different things to worry about between HR versus Procurement versus Legal).
Make sure it’s purpose-built for what you need; that is, use a tool that’s empowered by GPT, like the AI Front Door, rather than just letting everyone use GPT with no parameters or clear goals.
Education: Teach people in your organization how to use your tools properly. Some of that will be specific to your organization, and some will be part of the larger culture—for example, just as we all had to learn how to use an internet search engine, so too do we need to learn how to give LLMs prompts that will deliver the resolution we need. (That also makes it easier for your employees to follow your procedures and policies, which reduces risk.)
If necessary, gate the tool so people in your organization can’t accidentally use it destructively. (This is similar to the way we hold software licenses and permissions.)
Maintain strong data oversight, such that you know where your data is going and how it could be used—which you can check by looking up the Terms of Service on the tools you use, like OpenAI—and whether or not you can opt in or out of data collection.
The capabilities of AI, especially generative AI like ChatGPT, have become increasingly astonishing. But it’s instructive to take a step back and understand what AI really is and how it works—and what its limitations are.
Dig into this newsletter to learn more, starting with further reading below about the latest news around AI safety.
Top of the Ops
Beware of the folly of “mythologizing” new powerful tech like AI | New Yorker
Calling for a “Manhattan Project” for AI safety | Politico
Anthropic’s “constitutional AI” imbues AI systems with “values” so to make judgments about what it generates | TechCrunch
Speaking of Anthropic, Zoom invested in the company and will integrate its Claude AI into the video conferencing platform | Reuters
OpenAI CEO Sam Altman urged AI regulation when he testified before Congress | NYT
Recently, Tonkean announced LegalWorks, a matter lifecycle management solution that enables legal teams to harness the potential of AI-powered process automation technology safely—in a way that not only increases process adoption, improves performance, and creates business value, but that actively lowers risk and better ensures compliance. | Tonkean
If you have some insightful news or knowledge about the world of Ops to share, let us know!
Ops in the Field
AI is suddenly showing up in a big way in every vertical, market, and field, and each of those has its own set of challenges and opportunities when it comes to AI and safety. In Legal, the risks of AI like LLMs are perhaps obvious—that there’s a temptation to use LLMs like GPT-4 for researching or dispensing legal advice, for example. And also that employees will use tools like these in irresponsible ways such that they put their company’s or clients’ data at risk.
But there are huge potential benefits, too, like automating menial tasks such as legal matter intake. Another significant upside is making it easy for people to follow processes, which reduces risk and increases compliance and auditability. Some 67% of employees routinely skip legal processes altogether—a huge vector for risk.
LegalWorks is a matter lifecycle management solution that enables legal teams to harness the potential of AI-powered process automation technology safely—in a way that not only increases process adoption, improves performance, and creates business value, but that actively lowers risk and better ensures compliance.
And it’s equipped with Tonkean’s AI Front Door, an AI-powered intake experience accessible to employees via email, Slack/Teams, or a customized web portal that lets employees make requests of the legal team using common language prompts. For example, they can request an NDA by typing something like, “I need an NDA for ACME,” and Tonkean will automatically pull the account details from Salesforce and generate a new NDA doc via Docusign.
Another key component of LegalWorks is LegalGPT, which uses AI to enable legal teams to create processes that automatically triage and classify unstructured inbound business requests.
LegalWorks wraps seamlessly around whatever tools, policies, and systems your organization already has in place, which eliminates change management and lets people work within whatever apps and tools they already use. And it’s all entirely no-code, so operations teams can create all the automations they need without relying on IT or dev teams to do it for them.
To put it succinctly, LegalWorks was designed to finally give legal teams the technical means to do what modern legal departments need to do: create business value, save money, reduce risk, and increase process adoption. And it does so without ever jeopardizing your organization’s sensitive data or general security.
Tell us about clever solutions, success stories, and yes gift ideas you want your Ops peers to know about for future newsletters.
AdaptivOps is MOVING! This month, we’re very excited to announce the merging of our community, AdaptivOps, with Tonkean, to form The Tonkean Community!
What does that mean for you? Same community, same great content—including our podcast, events, and forum, but with more firepower behind it!
After three years of growing AdaptivOps, we’re pleased with how far the community’s come. As a result, Tonkean will be taking a more active role in managing it. Please know that this still means our community members will not be sold to. We remain a vendor-neutral space for all operations and automation professionals to share knowledge.
In this month’s Tonkean release notes, the biggest update is that Inner Items created in a parent module are now moved to a separate dedicated module. For example, if you have a module that handles the general intake for procurement requests, you may have a subprocess when a request requires special manager approval. You can create an inner item to handle this subprocess and add that inner item to its own module in the solution. We also have a ton of new and enhanced data sources, including Anthropic, AWS Textract, CaptivateIQ, Docebo, JIRA Service Management, Kallidus, Pingdom, and Whistic. As always, we fixed a few bugs, too.
Be sure to check out our newest initiative aimed at making Tonkean even easier to use, all on your own: Tonkean 101! Step-by-step guides and quick, snackable videos.
Check out the launch of our newest product, LegalWorks! It’s a matter lifecycle management solution that enables legal teams to harness the potential of AI-powered process automation technology safely—in a way that not only increases process adoption, improves performance, and creates business value, but that actively lowers risk and better ensures compliance. Go deeper into safety in AI and legal here!
Thanks for checking out Ops Digest! To learn more about who Tonkean is or what we do, we have a few different kinds of trials that you can sign up for. They walk users through our most salient solutions, including Legal intake, Purchasing Approval, Invoice Intake, and Email inbox automation. Sign up for one here!
Legal intake is inevitable, but managing it doesn’t have to be a slow, painstaking, manual process. With AI, automation, and intake orchestration, you can streamline intake processes and reclaim hours of time and effort.