text stringlengths 0 2.46k |
|---|
[1148.04 --> 1148.28] Right? |
[1148.54 --> 1148.78] Yeah. |
[1148.84 --> 1153.44] So you do have to off through Sourcegraph because that's how we, I mean, we wouldn't be able |
[1153.44 --> 1157.14] to provide it for free if you didn't off through Sourcegraph because on the back end, we're |
[1157.14 --> 1158.86] calling out to different language model providers. |
[1159.30 --> 1161.92] And we're also running a couple of our own. |
[1162.66 --> 1162.98] Okay. |
[1163.30 --> 1164.60] So accessible then. |
[1164.68 --> 1169.32] Not having to install Sourcegraph, have it scan my repository. |
[1169.32 --> 1173.92] Like the traditional way you provide intelligence, which is to leverage literally Sourcegraph |
[1173.92 --> 1174.78] on my repo. |
[1174.98 --> 1181.40] I can just simply off through Sourcegraph, have an extension in my VS Coder in the future |
[1181.40 --> 1181.96] Emacs. |
[1182.10 --> 1182.54] Exactly. |
[1182.90 --> 1183.58] Them potentially. |
[1183.88 --> 1185.14] They're kind of loosely coupled. |
[1185.36 --> 1189.08] Like we don't, we don't, we don't believe in strong coupling just for the sake of, you |
[1189.08 --> 1190.72] know, selling you more software. |
[1190.78 --> 1194.90] And I think with Cody, the design philosophy was like, look, if you connected to Sourcegraph, |
[1195.50 --> 1196.54] it does get a lot better. |
[1196.62 --> 1199.94] It's like if you gave a really smart person access to Google, they're going to be a lot smarter |
[1199.94 --> 1201.78] about answering your questions. |
[1202.14 --> 1202.20] Yeah. |
[1202.20 --> 1205.14] But if you don't give them Google, they're still a smart person. |
[1205.38 --> 1210.64] And so Cody will still fetch context from kind of like your local code using non-Sourcegraph |
[1210.64 --> 1213.10] mechanisms if you're just running it standalone. |
[1213.52 --> 1213.62] Yeah. |
[1213.82 --> 1216.38] How does it get this intelligence as an extension? |
[1216.52 --> 1218.30] Like how does that, explain how that works. |
[1218.34 --> 1220.28] Like I've got it on my local repo. |
[1220.40 --> 1220.90] It's an extension. |
[1221.06 --> 1222.58] How does it get the intelligence from my code base? |
[1223.06 --> 1223.30] Yeah. |
[1223.42 --> 1229.94] So it's basically, I mean, think of the way that you would like understand or build a mental |
[1229.94 --> 1232.22] model of what's going on in a code base as a human. |
[1232.94 --> 1235.44] You might, you know, search for some pieces of functionality. |
[1235.84 --> 1238.98] You might read through the readme, click on a couple search results. |
[1239.00 --> 1239.64] It does all that. |
[1239.72 --> 1240.84] It's reading my readme right away? |
[1241.28 --> 1242.22] Yeah, basically. |
[1242.22 --> 1249.10] So when you ask a question, Cody will ping Sourcegraph for, hey, what are the most relevant pieces |
[1249.10 --> 1252.02] of documentation or source code in your code base? |
[1252.14 --> 1255.68] And then essentially, you know, quote unquote, read them as a language model and use that |
[1255.68 --> 1256.98] as context for answering a question. |
[1257.38 --> 1260.26] So if you ask like a general purpose question, it'll typically read the readme. |
[1260.26 --> 1263.16] If you ask a more targeted question, like, oh, how do you do this? |
[1263.24 --> 1266.02] This, you know, one specific thing, like, you know, read a PDF or whatever. |
[1266.48 --> 1272.64] It'll go find the places in source code where you're, you know, it processes PDFs and read |
[1272.64 --> 1276.64] that in and then interpret that through the lens of answering your question. |
[1277.08 --> 1277.66] In real time. |
[1277.76 --> 1277.90] Yeah. |
[1278.12 --> 1278.24] Yeah. |
[1278.38 --> 1281.98] Is there a latency to the question to the gathering? |
[1282.28 --> 1283.48] And like, what's the speed? |
[1283.48 --> 1289.74] If I said that example, how does my application, you know, compile a PDF from a Markdown file, |
[1289.78 --> 1290.14] for example? |
[1290.38 --> 1290.54] Yeah. |
[1290.62 --> 1293.28] So it typically gets back to you within like one or two seconds. |
[1293.56 --> 1296.12] And most of the latency is actually just the language model latency. |
[1296.26 --> 1298.98] So it depends on what language model you're choosing to use underneath the hood. |
[1299.30 --> 1303.34] All the source graph stuff is super fast because that's just, I mean, there's no like, yeah, |
[1303.50 --> 1304.82] it's source graph is fast. |
[1304.90 --> 1306.88] We've spent the past 10 years making it very fast. |
[1306.88 --> 1312.80] And there's no like, you know, billions of linear algebra operations happening with source |
[1312.80 --> 1313.02] graph. |
[1313.10 --> 1319.24] Source graph is just, you know, classical, you know, CPU based, you know, code and text. |
[1319.48 --> 1319.90] What about privacy? |
[1319.90 --> 1320.18] Yeah. |
[1321.14 --> 1321.32] Yeah. |
[1321.40 --> 1325.92] So privacy is extremely important to us, both in terms of, you know, individual developers |
[1325.92 --> 1327.38] and our enterprise customers. |
[1327.38 --> 1331.68] Like the last thing they want to do is have their private code be used as training data |
[1331.68 --> 1335.38] into, you know, some general purpose model that's going to leak their sensitive IP to |
[1335.38 --> 1336.08] the rest of the world. |
[1336.08 --> 1343.08] So we basically negotiated zero retention policies with all our proprietary language model providers, |
[1343.28 --> 1347.68] which means that your data is never going to get used as training data for a model. |
[1347.88 --> 1354.26] And not only that, the language model providers will forget your data as soon as the request |
[1354.26 --> 1354.82] is complete. |
[1354.82 --> 1360.42] So like there, there is no persistence in terms of like, you know, remembering the code that |
[1360.42 --> 1364.80] you sent over to complete a request that just gets forgotten as soon as the language model |
[1364.80 --> 1366.60] generates a request for Cody. |
[1366.60 --> 1372.66] And then for the rest of it, I mean, Sourcegraph has always taken user privacy and code privacy |
[1372.66 --> 1373.32] very seriously. |
[1373.66 --> 1377.56] It's why we've been able to serve the sorts of enterprise customers that we do. |
[1377.76 --> 1377.98] For sure. |
[1377.98 --> 1381.54] I know why that's important, but why, spell it out, why is that important? |
[1381.82 --> 1385.56] What your, this zero attention policy, what's the real breakdown of that privacy? |
[1385.66 --> 1387.70] Why is it important to the main users? |
[1388.58 --> 1393.48] So from, from a company's point of view, it's important because you don't want to leak portions |
[1393.48 --> 1397.18] of your code base or have them persist in the logs of some third party data provider. |
[1397.78 --> 1402.16] As an individual developer, I think it's, it's just important to give you control over, |
[1402.34 --> 1403.24] over your own data. |
[1403.24 --> 1407.96] Um, and I think that's going to be an especially important thing, uh, in, in this new world |
[1407.96 --> 1413.98] that we're living in where, um, you know, before private data was, was valuable. |
[1413.98 --> 1415.52] Um, you know, it carries value. |
[1415.52 --> 1420.92] It tells you things about a certain person or the way they work and that can be used for, |
[1420.92 --> 1422.80] you know, purposes, both good and bad. |
[1423.02 --> 1423.74] Search history. |
[1424.36 --> 1425.50] It's like search history, right? |
[1425.50 --> 1425.90] Exactly. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.